There is a fascinating article titled "I’m a Luddite (and So Can You!)" by Tom Humberstone that, I think strikes at the core of this issue. In short, we aren't afraid of progress, we're afraid progress will be used against us. In the case of Hasbro, it's just another way to maximize shareholder value at the expense of customers and employees. We keep them from turning tools into weapons by being vigilant and vocal. I'm glad you're keeping an eye on them for us.
These are fair points, but not the common critique of AI up until a couple years ago. I remember when Andrew Yang was running in 2020 and people were literally scoffing at the idea of automation getting out of control. Just to be clear I did not vote for the guy, and I don't believe he's some genius or fortune teller. But if you were paying attention to what all the big tech companies were investing in back in 2019 and 2020, anyone could see the trend and predict this to happen. AI is already here, we missed the chance to get ahead of it. All we can do now is try to fix as much of the broken system as we can. Machine learning will never be outlawed, as it provides a massive benefit to the companies and countries that utilize it. Lawmakers surely know that to outlaw machine learning would mean to lose the tech race, and ultimately the US's power position on the world stage. We have to coexist in a post machine learning world, whatever that means... 🤷♂️ Anywho, no hate. Just trying to be real. Love to all my human brothers and sisters, struggling in these hard times. I'm always worried about what will happen next, and when my time will come. For now, I'm doing my best to try to learn skills that AI currently has trouble with, I suggest everyone do the same.
Hi Professor DM, professional code monkey and frequent professional user (but not programmer of) AI with a few Microsoft AI certifications. LMM stands for "Large Multimodal Model". It's pretty similar to what people think of when they think Large Language Model (LLM) but instead of just being able to take in text it can take in video, audio, and/or images as input data. Examples of LMM are Gemini, Clip, and Flamingo. It's not a term you see too often as most people lump them together with LLMs when talking about generative AI that's in the current hype cycle. Also, FWIW, I have dabbled a bit in using AI for my GMing, world building, and other RPG creation needs and found it severely lacking even though it's proved pretty useful at work. Except for some blocks of text I found a chore to start writing, I've almost entirely relied on good old fashioned human brain power for my TTRPG hobby.
good 'ol human creativity, and standard random generators always work best. i do not believe that will ever change. AI is more predictable (currently) than a human when coming up with stuff. But also not unpredictable enough compared to proper randomness for exciting random events.
I love how they can say "understanding the ethical ramifications of using AI in our work" but everybody knows that they really mean "how can we get away with it and avoid being sued". ^^
Suing someone for an AI model trained on their pictures is silly. Real people "train" their minds on what they see. Models for AI do not have the images, they have weighting which represents what they studied. Merge models, they do not increase in size. SO many people who complain have no idea what they are talking about. They are just worried their art jobs are going away, and their "special" talent isn't so special anymore. Let's get real. Dp we sue someone who uses a lick from a jazz musician in their solo? If so, every solo out there is a copyright violation. Same with all cosplay outfits, fan art... it is all silly.
@@ThePromptWizard2023 Except there's one slight problem, models aren't people. Humans take inspiration and create something new from that with elements of the old, most importantly humans can create things without the need for inspiration, it just helps. AI cannot do any of that, it recycles what it's fed and poorly at that, to create a highly uninspired, boring and regurgitated images. It is effectively worthless as a tool because it cannot think, which is also why it is terrible at anatomy, complex posing, perspective and lighting. The only real application of this technology currently demonstrated on mass is actively malicious. Artists are mad because their work is being taken without consent, regurgitated to produce broing, uninspired results and ultimately used to destroy their livelyhoods without even benifiting the consumers of their art. Comparing the neurological process of inspiration and the regurgitation of a souless computer is completely inhuman and speaks to a complete lack of creativity or experience in any kind of artistic field on the part of the individual making it.
To quote stand up comedian Robin Williams, when asked by a German interviewer why they don’t have comics like him in Germany; “as Germans did you ever consider you had killed all your funny people back in the 1930’s?”
I think most people know how useful the different AI technologies can be for artists, the problem is that the majority of corporations see the tool only as a way to get rid of the human element and cut costs (something that will backfire eventually if they can'f find some balance). As such, given the scummy behavior that companies have shown so far, I really can't blame people for being skeptical. I just hope that we come out of this transition period with a tool to enhance roleplaying and not a chatbot DM.
Backfire? Do you mean like how the push from hand drawn comics went to computer drawn comics? The human element is only needed until one of two things happens. 1. AI reaches the quality of the above average artist. 2. People accept the lower quality as "the new normal," this is more likely as all it takes is the companies all deciding AI is what they will use. Eventually there will be improvement but it won't be out of necessity it will be the result of AI updates improving the end product by random chance.
If you take company motivation out of the picture. Companies that make an equivalent product cheaper will put companies that don't, out of business. I don't like what AI will do to artist jobs, but logically I know from Econ 101 increasing use of generative AI is inevitable.
"With this character's death, the thread of prophecy is severed. Restore a saved game to restore the weave of fate, or persist in the doomed world you have created."
The Luddites get a bad rap. They were just protesting working conditions. They didn’t mind the machines. They smashed them because they were easy targets to make a point.
Indeed. It's later "upper class" people that used the term derogatorily to win the rhetoric battle. Luddites didn't want the machines because they took the jobs and Alienated from production process (they got treated as machines and used their human spirit to show the problem with dong this to a society.) if they were part of the actual business and not just labor to be replaced... Who wouldn't love using tech to do their job better? Sad day capitalist profit motive really just treats human life as a commodity.
For me the main issue is that WotC/Hasbro is a billion dollar company, they've shown willingness to cut corners with loads of their marketing material and even some of their recent books (Spelljammer in particular) I just can't trust that all they'll use AI for is generating simple concepts and expanding them. Companies love to take an inch and run with it way past the mile mark.
Hiring people with AI experience is not "to cut corners". AI is a tool like a word processor (and we don't regard using word over a type-writer as "to cut corners" do we?). Tools are there to help us do a better job, by simplifying tedious tasks (like rewriting a whole page because you wanted to move a paragraph from the bottom to the top). Bouncing ideas, tweaking some things to help us do a better job more efficiently - that is my view on AI.
True, but we don't hire people for the sole purpose of using a word processor. We hire people for specific tasks that happen to include the use of word processors. You hire an artist who knows how to incorporate LMMs as a tool, not rely on them for the entire product. You hire writers who know how to use LMMs to brainstorm ideas and double-check their formatting, not rely on them for the entire process. You hire someone who will incorporate LMMs as a way to increase their own productivity, not to become a priest to the Oracle of Delphi.
Essentially, the person doing the job needs to know more than the AI they're instructing does about the task requested. Hiring people who know more about the AI than they do the task they're assigned is cutting corners and results in lower quality work.
@@hweidigiv True, hiring someone to help the company to be more efficient is always a good thing, assuming you can afford it. Increasing the productivity as a creator is always welcome, and now we have another tool to help us be better us. I really like it.
oh please...companies will do ANYTHING to maximize profit. 'quality player experiences' is not Hasbro's primary concern regardless of what their PR statements, AI stance, or job descriptions say.
I'm not sure that's a counterpoint so much as the other side of the problem and the reason why people develop dangerous technology in the first place. Afterwards, the technology is used to help. We can hope that the transition is quick, we can work to keep the harm low, but the whole point is to get to the other side and make the technology safe.
I tried having AI make a one shot for me. It started off just fine: a city known for their prize winning show dogs had them stolen and held for ransom by a group of well organized bandits. The more I asked it the more it fell apart with changing backstories, characters being flip flopped, locations being completely ignored, etc. AI absolutely has its uses, but being a GM? It's not quite there yet.
As someone who has been training large language models (LLMs) as a job for the last year, I can definitely tell you that AI is only ever funny by accident. Try to get it to use puns, and it just puts random words in quotation marks. Try to get it to do something right, and it will do it so hilariously wrong that you can't help but laugh. To those saying that AI will someday be as smart as humans and able to DM games, I kind of doubt it, at least when it comes to LLMs. All those do is put words in sequence. There's no actual conceptual understanding behind the words they spit out, which becomes apparent when you start stretching the models in certain ways. AI is nothing more than a tool, and not one that I would rely on too heavily at this point.
I don't think that would stop a non-gamer CEO from trying to force it through, then trying to replace their game with that in a subscription model as was leaked to be their intention.
Artist here with my two cents: The problem I have with AI and art isn't necessarily the fact that its an automated process. It sounds like it can be a useful tool for artists as well as corporations. The problem lies in how those generative models have learned to produce their results; they have scraped and sampled from living artists who have not given permission, and have received no reimbursement, for their work to be fed into the machine. This is plagiarism. This is copyright infringement. This is theft. Unlike a person, who observes a master and learns their techniques, an algorithm is not sentient, and cannot innovate = only produce material in a mathematically determined way. I am all for development of an ethically sourced AI built from works within the public domain, but I can't in good conscience use these tools in my professional work if there's a chance of it effectively stealing work from other living creators. Thanks for attending my TED talk
"Unlike a person ... an algorithm ... cannot innovate = only produce material in a mathematically determined way" That is a too simplistic view on LLMs. They are not deterministic. The same prompt will give different answers. The prompt itself has a big influence on the result. A huge _issue_ with LLMs is that they "hallucinate" and "invent" facts (and present them as such). And what does "innovate" exactly mean anyway? You can still argue that the copyright holders never consented to that kind of use of their works. But the whole "innovate" line of argumentation just opens a can of philosophical worms IMO.
I too am an artist and agree with you, but I'd go one step further and suggest that AI is not a tool when it comes to artwork, it does the work for you by following your prompts. For example, a broom is a tool, you use it to clean your floor. Your house cleaner is not a tool, the cleaner does the work, following your prompts. Anyway, that's how I see AI and I've heard the, "it's just another tool, like a pencil or ruler," argument plenty of times and I don't think it holds water.
You make a good point, but I would argue that it is not plagiarism/copyright infringement because the ai takes the techniques and concepts, just like an art student, and applies them. It's not like you can charge an art student for looking at art on the internet.
@@andrewtramel4390 The art on the internet (theoretically) is there either because it has been commissioned and paid for by someone, or it is being used by the artist to promote their services- in other words to generate work for themselves. It's not put there for free, and the fact that it is available for free doesn't imply that it is there to be taken and used by other people in their commercial work for free. Of course someone can easily take the work and they can learn from it, but they can't (in theory) use it commercially. Even if a student, or let's say 100,000 students use the work to learn from, it is unlikely they will be any professional threat to the original artist whose reputation improves with the more humans who look at, study, share and emulate that artist's work. This leads to more work for the initial artist. Over the last decade, artists have worked very hard to build these kinds of audiences on social media for the exact reason that it benefits their career in numerous ways. The price the artist pays for that audience is giving their work away 'for free'. This is not the same with AI. When AI ingests an artist's work, the people who created that AI system are benefiting commercially from the use of that art. The ingested artwork increases the value of their product, usually with no compensation paid to the artist and no consent. The artist is not credited for their work. The artist gains no reputational benefits for the use of their work. The artist gets absolutely nothing. When a human artist learns from another human artist, the student can still only work human hours. There is a human limit to the amount of competition that the student represents. When an AI ingests an artist's work, it can pump out images that replicate the original artist's style day and night, at an inhuman rate. Meanwhile, that AI generated imagery is being used commercially by entities that formerly would have hired that original artist, but now they can get something very similar to that artists work and use it commercially and pay nothing, or very close to nothing for it. You can split hairs about whether this is plagiarism or not, but the end result is the same: the artist's work is being used without their knowledge or consent, it is being used without any reputational benefit to the artist, it is putting the artist out of work and people other than the artist are making money from the process. Whatever you want to call it, it's as bad as plagiarism or worse.
I was going to point out that it was far from veiled sarcasm, since it was out there shouting “look at me!” And then I realized that I overlooked sarcasm…and my GenX soul is now punishing me for this transgression by forcing me to watch RUclips kids with my little.
This is a natural step. AI is a tool that we should use more. A lot of people seems to think that AI is used as an auto-pilot, when it should be regarded as a co-pilot. Good used is to summarize, filing, and fact checking. I would embrace it to the fullest and is a little surprised that they didn't do it earlier.
AI has already replaced most low level writers who were doing articles at $5 each. I worked as a managing editor with a 100+ writing team focused on high quality work. We could tell the difference when a writer tried to use AI (it just didn't meet our quality standards and we had very high standards). BUT the reality is AI creates better quality writing than 95% of writers. If you're a writer you really need to work on your skills so you're in the 5%. At that level AI doesn't come close to competing. I suspect the same is true in most other creative fields where AI is being used. I have found AI useful for brainstorming ideas. It almost never comes up with an idea I use but it gives you a whole pile of different ideas that get your imagination working so you can come up with something original and out of the box yourself.
Me, speaking to the A.I. GM: "I find your lack of imagination disturbing." You are correct when you say A.I. is just a tool, and I would venture to say you, like I, would not look to a tool for inspiration. I shall be the inspiration, and the tool shall assist me in bringing it to fruition; not the other way around as seems to be the trend in movies these days.
I'm the opposite; I'm terrible at the inspiration part. I can take a bad module and make it good, but I can't take no module and create anything. I can use generative AI to build bad modules that I edit into good modules.
To answer your question: a LMM stands for 'Large Multimodal Model' - it basicaly means a model that can deal with different types of data such as text, audio, video, images etc all at once. to contrast ´chat gtp for example is based on a Large Language Model that only deals with text.
Anybody who ever watched Star Trek the Next Generation will remember what would happen whenever Data would try to tell a joke, that's how I see AI's attempt at humor going.
NGL, I'm morbidly curious to see how AI would handle a failed sanity check from a heavy Far Realms encounter. Would the madness spread? "Welp, there goes humanity!"
It is true that the Industrial Age created new jobs to replace those jobs lost by the transition from the Agrarian Age, but those jobs were detrimental to health, created overcrowded cities, destitution, hunger and a raft of other issues such as alcoholism and petty crime. When people lived and worked in the agrarian world, their jobs were healthier, and food was plentiful.
Imagine for a moment if someone came to your home and took all the clothes out of your closet. They tossed those clothes into a wood chipper. They picked up the shreds and dipped them into fabric glue and twisted them round into giant wads that looked vaguely like shirts and pants, and they charged you for these "new" clothes. That's the current model of AI. I don't even like using the term nexus at some point, we ARE going to create real AI, and we'll have to call it something stupid and obtuse because the term Artificial Intelligence will be tainted and politicized.
Yeah, AI has been in our tools for a couple decades by now. Photoshop, Maya, game engines, word processors, etc. I think most people are objecting to generative AI, largely because it's doing something AI hasn't really done in the past.
LMM: Large Multimodal Model. It's a multipurpose AI. It can handle everything, text, pictures, sounds, and video. An example is Google Gemini. This is opposed to an LLM (Large Language Model) which deals only with understanding and generating human language (ChatGPT, for example).
Thank you for your balanced take on the subject. The doom sayers for A.I are really putting a damper on what could be a fun cool tool if they just embraced it with caution.
I think that like everything else, DM's will ultimately utilize the tools that interest them & supplement their particular style of play. When I first played D&D in the late 70's, my DM's were "typical nerds" (intelligent, creative, socially awkward) and older than us. As they graduated high school or moved on, I, being the 'next-nerdiest' of my friends, became the full-time DM running the classic modules. Soon I was stringing these together in increasingly complex story arcs and substituting more and more content--what we now call homebrew. To your excellent point, Professor DM, this was an adaptive response to my player's input: our collective style. I do not believe that AI can replicate that specific type of synergy--at least not yet. As any DM worth his dice will tell you, go ahead and "borrow" (and adapt) any tool out there to achieve the results that best meet your table's desires. Keep on rollin'!
In my Chat GPT experiences (limited though they have been), I've found three things to be generalizable statements: 1) Asking for factuality demonstrates the creativity of the engine, because it outright makes things up (including references to non-existent articles/books!) but they *look/sound right* to the untrained/uncritical eye/ear. 2) Asking for narrative, though, is pretty awful, and it can get very nitpicky on particular characters and attested lore/sources (it would be good at the quiz show "Um, Actually"!), but then asking for poetry also generates doggerel that may rhyme fairly well, but is boring and cliche. 3) Something that it does quite well, though, and can save some time and agonizing on for people like me--who love to write and do it well, but dread writing certain things--is cover letters and short business e-mails. They can end up being a bit jargon-y and use a lot of buzzwords (which, unfortunately, HR people and business types love!), and of course they need to be read over and checked and certain small things generally need to be altered or expanded, but this is a huge time-saver. Thus, I suspect it might be able to describe certain things that a DM might find useful if they're not sure what to say about certain rooms and such, but creating a cohesive narrative is probably beyond it at this stage. AI image generation, though, is wonderful as far as it goes, especially for those of us who aren't great with art, and if one is only using it privately and not in a publication, I don't think there's anything at all wrong with using it; that being understood, though, not all programs/websites/apps in that regard are equal, and some are far better than others (including amongst the ones available for free), and the results on these can be surprisingly good and creative, while also being occasionally disastrously awful but good fodder for laughs (even independent of the tendencies to have super/subnumerary fingers and toes, or even whole limbs, etc.). I recommend one called DezGo if you're interested in experimenting with it (dezgo.com); it can do Elves and humans/humanoids well; anything else is a bit of a crapshoot. ;)
Having played around with Artbots and knowing how to draw myself, trust me, you're better off spending the time learning how to draw. There's a lot of videos here on RUclips about getting started. I'd go with Marc Brunet's video since he goes step by step on how to do it. Search "marc brunet 30 days" and it should pop up as the first video.
Are you, perhaps, suggesting there is a SPIRIT to the art of role-playing that a machine cannot possibly replicate? Such materialist language, I wonder to what degree intentional. Great story. Love your coverage.
Hahahaha the 'Definition of A Joke' was a master class moment! Had to relisten to it, it was so good 😂 The intense level of dryness in the delivery*chef's kiss*
Remember when artists in the 80s and 90s were big names. Frazetta, Elmore, etc. I mean Image comics artists even had commercials on TV for levi buttonfly jeans! Same with Hollywood, I dont think the names carry so much wait as they used to with say Bruce Willis, Sigourney Weaver, Linda Hamilton, Arnold Schwarzenegger. I think it is related. I am not sure what to call this phenomenon but something is definitely different. It used to be the stars that pulled us, but now that everything is a sequel, prequel or remake, we just follow a IP instead of a artist
Also the sheer nature of the world we live in makes these identities very VERY culturally isolated. It's not that we don't have "movie stars" anymore... It's just that there are smaller and smaller pockets of micro stars. We don't ALL watch TV shows at the same time on limited available broadcast schedules. and if we wanted to watch the new movie we ALL went to the theater... Now all of that and more is at a finger tips. People are just more isolated cultural in some ways. And yah some of that is by design of the corps.
I use grok for hooks from specific settings prompts, monsters specific to areas, spellbook generating for wizards and dragons that are based on a theme for the wizard or dragon.
In other words, Robo-DMs. I think this is inevitable. It doesn't matter whether humans are better. There aren't enough of them. For the VTT to succeed at volume, they probably need robo-DMs. The unpaid human DM is the weak link here. Also, I'm sure we all agree that fully detailed 3D VTT maps are too expensive to be created on the fly. A robo-DM could do that. This will prevail for economic reasons -- it is way cheaper; no one wants to pay for the alternative -- so for it to exist, the robo-DM will be a thing.
I fed the descriptions of 20 rooms into AI and asked it to assign a difficulty factor for each room (you would call it a room dc ) and in 2 seconds it gave me both the DF and an explanation as to why the room was assigned the DF. This saved me an hour or two of my time that I could use to create game mechanics I needed for the special encounters. I love AI as a tool. They Just announced AI chips (like physics engine chips) to go into new PC going forward. For Video games it will soon be used to generate areas that the player wanders into but the DEV did not have time to make. "I for one welcome our AI overlords".
One thing doesn't pass the smell test, Professor; "video games, down the road" is not a real job position. Everything on the posting suggested utilizing and "road-mapping" AI sooner, not later.
Hi Professor, excellent video. LLM stands for large language model, aka the type of AI algorithm that Chat-GPT and stuff like Midjourney are based off. It's a technical term, not out of place in a job posting. This is in order to differentiate it from other AI models (from data science prediction algorithms, to Chess players, to general AI).
AI is veeery useful for creating Minor npcs and to do research for creating random Encounter Tables etc. I think the first step would be GM tools for the vtt
I work in the video game industry, we've started using AI tools to help with game development and QA, I've had a chance to work with the tools and the teams indirectly and I can confirm the description of what they're asking for sounds a lot like the job description of the engineers I've worked with, it sounds like Hasbro's game development interests are just trying to catch up with what the big video game studios are already doing.
Wizards has been talking about an AI dungeon master since 4th edition and the precursor to dnd beyond. It will never be a replacement experience, but may be able to faithfully run groups through premade modules (perhaps stilted, without a special DM flair). I think this is fine. No replacement for the real thing, but for example I would love a tool that runs me through some old adventures my group has no interest in playing.
Jokes died in 2017… the same year your oldest video still on RUclips was published… Just kidding! I’m a huge fan of the channel, and not just your drama videos! The campaign recaps, XDM recommendation, and class reworks have all been super inspiring for my game. Thanks a bunch for your videos!
I like how being able to speak about how using AI is ethical is part of the job description as one of the elements behind being hired to work. I think it's because there is so much fan backlash against AI generated content that professionals have to justify their jobs to the fan community. This is disappointing because in my opinion the fans need to calm down about AI generated content. What matters is not whether a few precious jobs are saved (protectionism just doesn't work) but rather whether the content being produced is any good. We should be grateful that AI will allow more quantity if not quality of content being produced so that we can have more to consume.
I agree with this making sense for the VTT. I could see this possibly being for tables that want to play D&D but none of the players want to DM. Based on playing around with current language models, I agree that the tech isn't ready to take on the seat of DM, and I'm not sure if it will be any time soon. I could see it working as a way to make NPCs more interactive in the VTT.
Years ago I was in a sculpting competition and was up against the digital printing stuff. It felt like going to a weight lifting competition against a fork lift. The digital artists mocked us with "Art is art, it does not matter were it comes from!" Now that they are in the same position against AI, it gives me a chuckle, but I'm still not a fan of AI.
As someone who is rather fond of AI and use it for hobby stuff and messes with it as a hobby ‘games’ using AI as a major part of the experience mostly as a co-writer (imagine a TTRPG in which the players and the DM share the same jobs) of it already exist. AI Dungeon was kind of the front runner of that with effectively advertising itself as infinite text adventures through a LLM with a few others coming up over the years. These however to this day still have more than a few hang ups. Namely they have got to be handheld or they go off into weird tangents/break from logic and if you don’t shackle the AI pretty heavily they tend to get really naughty in to what they will generate. But censuring the models usually just makes them bad at everything as even perfectly mundane inputs cause it to freak out and stop. I wouldn’t trust AI to act as a DM on its own unless you are up for a particularly trippy experience as dead NPCs are just casually back, the DM constantly telling you that you randomly die, and a adventure that will regularly try and devolve into smutt at its nearest convince.
ChatGPT Prompt: Write an original joke about a RUclipsr with a channel named Dungeon Craft and make sure the joke isn't just a derivative of obvious and well known jokes. ChatGPT response: Why did the Dungeon Craft RUclipsr refuse to play Monopoly? Because he said if he wanted to spend hours arguing over imaginary property, he'd just read the comments section on his latest dungeon build video! I don't know... AI seems to be on par with the creativity of most RUclipsrs 😂
So, right now we see D&D as being a social experience where we're coming together to hang out and have fun with friends, playing a TTRPG. The experience is communal and flexible, with just enough rules to identify the boundaries and stakes, creating a sweet spot where the magic happens. That version of D&D as we know it, I feel, is safe from AI. I don't think you can replace that. Spontaneity, personality, expectations... an AI could be 'fair' and do what we want, but I feel it would create a more video game experience, since the AI will always have less agency than the live player. I don't think many players would want a table where an AI has more agency than they do. But we can trust, know, and grant that position to a live person. And once AI can host immersive experiences, I suspect we'll see many of the other trappings and boons of video games and even VR coming along with it. Considering that, I think AI will create a new space for TTRPG that exists between live D&D and video games. Single-player adventures with increased player agency and freedom, or very highly-trained modules. Even if players have to accept that if they go too far outside of the boundaries of its training data, it may have to tell them no, the same way a live DM might tell a player no to something they have no prep for, if it creates an issue for them running the campaign. That may well be an acceptable tradeoff for a currently unrepresented demographic of potential players who have difficult or limited schedules or other accessibility issues that create a barrier for playing live D&D. AI-run TTRPG could also be a great learning tool for new players to test the waters, or try out new systems. Imagine being able to go through a solo adventure, being given advice and guidance tailored to your play and knowledge. Getting to familiarize yourself on your own schedule and time, doing whatever you want, could give players the confidence and readiness to hit the game running when they have limited time to play live games with friends. I'm fully prepared for companies like Hasbro to make horrible products and try to use them to reduce overhead and charge players extra for less. "We made this shoddy campaign, but you get access to an AI companion that will patch it however you like! Probably!" But there's a lot of amateur innovation with AI that's surprisingly accessible, so I fully expect that fan projects to make a GOOD AI DM will produce compelling results before long. And as a more approachable expectation, will also create AI companions for DMs that can help them with campaigns, generating maps quickly, creating custom art, writing songs and speeches and descriptions for otherwise-rote stuff or patching in to libraries of artist-created assets to customize them for your use. We face a lot of ethical challenges with AI, and companies will try to bend it to both cynical and immoral spaces. But I honestly believe in the community and creative power of the TTRPG crowds to get involved, stay vocal, and innovate uses that will support the hobby instead of damaging it.
I'm known of the Twitch channels i hang out in, for Dad Jokes and Puns! : ) I once sent a list of my top 10 favorite puns to 10 of my favorite friends, and asked them to vote on their favorite pun... after tallying all the votes, to my dismay, no pun in 10 did
They have given up on lying about AI in their products. They’ve likely been using AI to write adventure paths and other source books as well. I have long since stopped supporting WotC (since 2017) and I won’t purchase anything labeled “For Use with 5E” or similar. Let WizBro’s TTRPG die on the vine. Support OSR & Indy that has divorced itself from 5E.
I like Chat GPT for generating game ideas, and exploring a few different NPC concepts. If I know I want part of a game set in a swamp, but I'm not sure what I want for the challenge/bad guy/adventure hook, it's very much like a random adventure roll. That said, considering the amount of fresh water it takes to operate the machinery, I'm not sure it's worth it. It also has canned phrases that crop up over & over, so anyone using it a lot is going to end up with that voice permeating their own adventures. And I think I've exhausted it's store of riddles (which are not very good).
The advent of photography is a really interesting technology with parallels to modern day AI. Originally it began relegated as a research tool, before the pictorialist movement elevated it to an accepted art form. People feared it would replace painting, claimed that it would ruin livelihoods, and insisted photography couldn't possibly be art. The reality is photography changed things, sure. The world adapted to the new accessibility photography brought - but life carried on and progressed. These tools both changed the accessibility of image generation, and from an economic viewpoint you can't increase something's abundance without decreasing it's value. In my experience, many people who dislike AI inherently dislike how this ease of access implies a devaluation of existing works, even when they find it hard to put into words. However, people adapt and end uses change to fill niches. Photography has since developed to a level nobody back then might have dreamt of, and AI may well develop on a similar trajectory in the future. However, the box is open and all we can do now is adapt. I'm always wary of saying AI could NEVER though, rather than AI currently can't!
You are right about keeping all the AI machines running. They showed a relaxing future of it in the Matrix. Chaplin's 1936 movie showed the bright future brought about by the industrial revolution also. Luckily the end of the agrarian age didn't do horrible things like raise pollution or excellorated climate change. Nope, all smooth sailing on the AI boat.
AI is unlikely to get better as a tool. It might have already peaked. The problem is diminishing returns, and the bottleneck of training data. All the training data used in image generators like Dall-E and Midjourney were obtained by scraping sites like Art Station and Twitter. That was the low-hanging fruit, though. First off, there is a cultural bias in the data samples and in the people collecting the samples, and it is reflected in the style of images AI can generate; diversifying the training data is going to take more legwork. Second, artists don't like their work being used in training data without their permission, and they will do everything they can to prevent it. One tactic is using programs like Nightshade to poison AI's with patterns hidden within your artwork's pixels. Third, the proliferation of AI generated images on the internet means that the potential pool of training data is now contaminated. The kind of web scraping used before will result in AI's being trained on AI generated images; this may create a feedback loop, or reinforce the messed up hands problem. All in all, AI generated images are only suitable for cheap slop. I would expect as much from WotC. I will continue to boycott their products, and I will continue to support my favorite artists by buying their books.
I understand why professional artists are worried about AI, but my experiences at the table have been very positive. My GM loaded his preferred random tables into an AI, and it is able to instantaneously produce a very complete profile of an NPC or lacation , including an image in a second. It's not publishable quality art, but it is more than sufficient for play.
I like this channel more when it focuses on the hobby itself. Really not thrilled with how much kickstarter promotion and hasbro news ProfDM is covering lately.
I just enjoy writing stuff and AI feels like I’d be cheating myself of something I enjoy. But I feel that way about something like The Fantasy Adventure Builder or pre-generated outlines, too.
I work with AI daily, and it has been advancing rapidly. Currently my team is focused on observability, validity, and proper Prompting techniques. We have already replaced a few traditional ML models with 3.5-turbo, and 4o is going to be replacing a yolov5 model we use. Basically I can see the technology replacing even more then can today in just a few years. As a DM it still has trouble building dungeons, quests, and adventures, however I have trained a model using my campaign's notes and its become my perfect Co-DM. Whenever I can't remember a detail or quest line from years ago it can find it for me.
Here is an original joke from ChatGPT: “Sure, here's an original joke for you: Why did the scarecrow become a successful motivational speaker?Because he was outstanding in his field!”
To me, the excitement of AI in dungeons and dragons is how it can be used with the NPC‘s. I envision games on a digital level, becoming more and more self supporting, and the players will enter into these environments. For example, let’s say a village is created with a bunch of NPC‘s, and the AI is going to be working the NPC‘s. So this village would be almost like a living village or a village with real people and when the players come in, they’re going to be able to interact with these people on a more detailed level. It’s almost like a sandbox environmentthat heals itself up and keeps itself going, and players are visitors.
Thank you for the thoughtful, balanced analysis! I think there's a bit of a moral panic about AI in some corners of the TTRPG hobby. I tend to think that AI will be harnessed to augment human creativity rather than replace it. The rumors of human DM extinction are greatly exagerated!
AI will probably be able to replace humans in some jobs, but it’s got a long way to go before it can do that. I think we will probably start seeing it in our lifetime, but it will probably be future generations who will really improve it.
AI DMs would be like open world video games like the Witcher 3 or Elden Ring. To be able to adlib on the fly outside of your basic video game limitations is still well into the future, at least enough to satisfy gamers.
until the program actually understands what it is saying or being told, it'll be very jank and ultimately worse than a human GM that's on top of removing human interaction from a tabletop game that is effectively built off of human interaction, which isn't a very smart idea since it will effectively gut the core of what ttrpgs are about in the first place.
what makes open world RPGs feel good is that everything you find has been placed by someone, every funny or interesting dialogue has been written by someone Generative AI is incredibly dull and uninteresting
I've found AI very useful for generating a dozen ideas for something. Usually most of the ideas are just bad or fail to take into account the circumstances of the situation. Sometimes there are no good ideas in what the AI produces, but even in those cases, it will almost always inspire a good idea in my head. Pretty much all of my players are using AI to generate portraits of their characters, and I've been using it to create portraits of NPCs and landscapes to help my online players see what I'm talking about. It's not great, but it is affordable, where my own "staff" artist wouldn't be. And it's pretty good at taking a sketch I've made and taking it 90% of the way to finished, where I can do the final touch-ups with GIMP.
I don't understand why people talk about AI game masters as if that would be the only way of using AI in ttRPGs. Surely to make a fully independent GM is not quite possible with the current tech (we will get there one day though), but AI tools could be a great help for GMs while running the game. For example an AI could listens the game on the background. Then GM could any time ask help from it. For example it could ask AI to invent an interesting twist to the story or to fast create a suitable NPC for a certain situations. Or AI could even totally take over some background events: what are the 10 most central NPCs or factions doing while the PC are busy with their things, and how that could become visible to the PCs.
"If you hate me, there's a stat block, you can kill me!" 😂😂😂😂😂
😂😂😂😂😂
_Old McWizard's had a firm_
_AI, AI, Ohhhhh_
Good one!
@@DUNGEONCRAFT1 Ta :)
@@DUNGEONCRAFT1 Please make Death Bringer sing that in the next one.
They sort of have to, after firing their entire book team. That's their fault.
It’s as if it was by design…🤔
That was part of the plan. This company deserves to be dismantled.
There is a fascinating article titled "I’m a Luddite (and So Can You!)" by Tom Humberstone that, I think strikes at the core of this issue. In short, we aren't afraid of progress, we're afraid progress will be used against us. In the case of Hasbro, it's just another way to maximize shareholder value at the expense of customers and employees. We keep them from turning tools into weapons by being vigilant and vocal. I'm glad you're keeping an eye on them for us.
Thank you. Nobody seems to know what the Luddites actually believed, they just banty about the term like a synonym for the Amish or something.
Is it available in PDF form?
These are fair points, but not the common critique of AI up until a couple years ago. I remember when Andrew Yang was running in 2020 and people were literally scoffing at the idea of automation getting out of control. Just to be clear I did not vote for the guy, and I don't believe he's some genius or fortune teller. But if you were paying attention to what all the big tech companies were investing in back in 2019 and 2020, anyone could see the trend and predict this to happen.
AI is already here, we missed the chance to get ahead of it. All we can do now is try to fix as much of the broken system as we can. Machine learning will never be outlawed, as it provides a massive benefit to the companies and countries that utilize it. Lawmakers surely know that to outlaw machine learning would mean to lose the tech race, and ultimately the US's power position on the world stage. We have to coexist in a post machine learning world, whatever that means... 🤷♂️
Anywho, no hate. Just trying to be real. Love to all my human brothers and sisters, struggling in these hard times. I'm always worried about what will happen next, and when my time will come. For now, I'm doing my best to try to learn skills that AI currently has trouble with, I suggest everyone do the same.
Where do I sign up for the Butlerian Jihad?
@@PsychedelicLasagna"Nukes are already here, so we might as well start WW3!"
Hi Professor DM, professional code monkey and frequent professional user (but not programmer of) AI with a few Microsoft AI certifications. LMM stands for "Large Multimodal Model". It's pretty similar to what people think of when they think Large Language Model (LLM) but instead of just being able to take in text it can take in video, audio, and/or images as input data. Examples of LMM are Gemini, Clip, and Flamingo. It's not a term you see too often as most people lump them together with LLMs when talking about generative AI that's in the current hype cycle.
Also, FWIW, I have dabbled a bit in using AI for my GMing, world building, and other RPG creation needs and found it severely lacking even though it's proved pretty useful at work. Except for some blocks of text I found a chore to start writing, I've almost entirely relied on good old fashioned human brain power for my TTRPG hobby.
Thanks for sharing your knowledge!
good 'ol human creativity, and standard random generators always work best.
i do not believe that will ever change.
AI is more predictable (currently) than a human when coming up with stuff.
But also not unpredictable enough compared to proper randomness for exciting random events.
I love how they can say "understanding the ethical ramifications of using AI in our work" but everybody knows that they really mean "how can we get away with it and avoid being sued". ^^
Suing someone for an AI model trained on their pictures is silly. Real people "train" their minds on what they see. Models for AI do not have the images, they have weighting which represents what they studied. Merge models, they do not increase in size. SO many people who complain have no idea what they are talking about. They are just worried their art jobs are going away, and their "special" talent isn't so special anymore. Let's get real. Dp we sue someone who uses a lick from a jazz musician in their solo? If so, every solo out there is a copyright violation. Same with all cosplay outfits, fan art... it is all silly.
@@ThePromptWizard2023 Except there's one slight problem, models aren't people. Humans take inspiration and create something new from that with elements of the old, most importantly humans can create things without the need for inspiration, it just helps. AI cannot do any of that, it recycles what it's fed and poorly at that, to create a highly uninspired, boring and regurgitated images. It is effectively worthless as a tool because it cannot think, which is also why it is terrible at anatomy, complex posing, perspective and lighting. The only real application of this technology currently demonstrated on mass is actively malicious. Artists are mad because their work is being taken without consent, regurgitated to produce broing, uninspired results and ultimately used to destroy their livelyhoods without even benifiting the consumers of their art. Comparing the neurological process of inspiration and the regurgitation of a souless computer is completely inhuman and speaks to a complete lack of creativity or experience in any kind of artistic field on the part of the individual making it.
Heh. As a German, I always knew this "joke" thing wasn't going to work out in the long run.
To quote stand up comedian Robin Williams, when asked by a German interviewer why they don’t have comics like him in Germany; “as Germans did you ever consider you had killed all your funny people back in the 1930’s?”
The joke about jokes was hilarious. I'm gonna have to borrow it.
Professor DM performed a vital public service by explaining the mechanics of a joke.
Especially aimed at Milliniels and Zoomers, heh heh
I think most people know how useful the different AI technologies can be for artists, the problem is that the majority of corporations see the tool only as a way to get rid of the human element and cut costs (something that will backfire eventually if they can'f find some balance). As such, given the scummy behavior that companies have shown so far, I really can't blame people for being skeptical. I just hope that we come out of this transition period with a tool to enhance roleplaying and not a chatbot DM.
AI requires the human element. We are smarter than corporations.
Indeed, true. And that's human nature. A balance must be struck. The problem is that lots of people will suffer until that balance is found.
Backfire? Do you mean like how the push from hand drawn comics went to computer drawn comics?
The human element is only needed until one of two things happens.
1. AI reaches the quality of the above average artist.
2. People accept the lower quality as "the new normal," this is more likely as all it takes is the companies all deciding AI is what they will use. Eventually there will be improvement but it won't be out of necessity it will be the result of AI updates improving the end product by random chance.
If you take company motivation out of the picture. Companies that make an equivalent product cheaper will put companies that don't, out of business. I don't like what AI will do to artist jobs, but logically I know from Econ 101 increasing use of generative AI is inevitable.
"With this character's death, the thread of prophecy is severed. Restore a saved game to restore the weave of fate, or persist in the doomed world you have created."
“MOURNHOLD! City of Light, City of MAGIC!"
The Luddites get a bad rap. They were just protesting working conditions. They didn’t mind the machines. They smashed them because they were easy targets to make a point.
Good point.
Indeed. It's later "upper class" people that used the term derogatorily to win the rhetoric battle. Luddites didn't want the machines because they took the jobs and Alienated from production process (they got treated as machines and used their human spirit to show the problem with dong this to a society.) if they were part of the actual business and not just labor to be replaced... Who wouldn't love using tech to do their job better? Sad day capitalist profit motive really just treats human life as a commodity.
For me the main issue is that WotC/Hasbro is a billion dollar company, they've shown willingness to cut corners with loads of their marketing material and even some of their recent books (Spelljammer in particular) I just can't trust that all they'll use AI for is generating simple concepts and expanding them. Companies love to take an inch and run with it way past the mile mark.
Hiring people with AI experience is not "to cut corners". AI is a tool like a word processor (and we don't regard using word over a type-writer as "to cut corners" do we?). Tools are there to help us do a better job, by simplifying tedious tasks (like rewriting a whole page because you wanted to move a paragraph from the bottom to the top). Bouncing ideas, tweaking some things to help us do a better job more efficiently - that is my view on AI.
True, but we don't hire people for the sole purpose of using a word processor. We hire people for specific tasks that happen to include the use of word processors. You hire an artist who knows how to incorporate LMMs as a tool, not rely on them for the entire product. You hire writers who know how to use LMMs to brainstorm ideas and double-check their formatting, not rely on them for the entire process. You hire someone who will incorporate LMMs as a way to increase their own productivity, not to become a priest to the Oracle of Delphi.
Essentially, the person doing the job needs to know more than the AI they're instructing does about the task requested. Hiring people who know more about the AI than they do the task they're assigned is cutting corners and results in lower quality work.
@@hweidigiv True, hiring someone to help the company to be more efficient is always a good thing, assuming you can afford it. Increasing the productivity as a creator is always welcome, and now we have another tool to help us be better us. I really like it.
oh please...companies will do ANYTHING to maximize profit. 'quality player experiences' is not Hasbro's primary concern regardless of what their PR statements, AI stance, or job descriptions say.
I love all of the Professor's dad jokes.
AI of Vecna, that is chefs kiss funny.
Knowledge of what a good joke is: I'm a dad.
Fool around and find out,.
When does a joke become a dad joke? When it becomes a parent...
Explaining what a joke is got a legit lol from me. 😂
Thanks for watching!
Same, haha
Was that one of those joke things?
The irony is they wouldn't get that it was a joke itself! Genius!!
Technology will always be used to harm people before it will be used to help them.
I am undergoing treatment for a serious condition right now. It's helping me live.
I'm not sure that's a counterpoint so much as the other side of the problem and the reason why people develop dangerous technology in the first place.
Afterwards, the technology is used to help. We can hope that the transition is quick, we can work to keep the harm low, but the whole point is to get to the other side and make the technology safe.
That's clever and wrong.
I tried having AI make a one shot for me. It started off just fine: a city known for their prize winning show dogs had them stolen and held for ransom by a group of well organized bandits. The more I asked it the more it fell apart with changing backstories, characters being flip flopped, locations being completely ignored, etc.
AI absolutely has its uses, but being a GM? It's not quite there yet.
"But what does this all mean for you and your table?"
Nothing. It means nothing
Lol. Maybe not.
Nor mine: I play BECMI and WEG Star Wars.
As someone who has been training large language models (LLMs) as a job for the last year, I can definitely tell you that AI is only ever funny by accident. Try to get it to use puns, and it just puts random words in quotation marks. Try to get it to do something right, and it will do it so hilariously wrong that you can't help but laugh.
To those saying that AI will someday be as smart as humans and able to DM games, I kind of doubt it, at least when it comes to LLMs. All those do is put words in sequence. There's no actual conceptual understanding behind the words they spit out, which becomes apparent when you start stretching the models in certain ways.
AI is nothing more than a tool, and not one that I would rely on too heavily at this point.
I don't think that would stop a non-gamer CEO from trying to force it through, then trying to replace their game with that in a subscription model as was leaked to be their intention.
Artist here with my two cents:
The problem I have with AI and art isn't necessarily the fact that its an automated process. It sounds like it can be a useful tool for artists as well as corporations.
The problem lies in how those generative models have learned to produce their results; they have scraped and sampled from living artists who have not given permission, and have received no reimbursement, for their work to be fed into the machine.
This is plagiarism. This is copyright infringement. This is theft. Unlike a person, who observes a master and learns their techniques, an algorithm is not sentient, and cannot innovate = only produce material in a mathematically determined way.
I am all for development of an ethically sourced AI built from works within the public domain, but I can't in good conscience use these tools in my professional work if there's a chance of it effectively stealing work from other living creators.
Thanks for attending my TED talk
"Unlike a person ... an algorithm ... cannot innovate = only produce material in a mathematically determined way" That is a too simplistic view on LLMs. They are not deterministic. The same prompt will give different answers. The prompt itself has a big influence on the result. A huge _issue_ with LLMs is that they "hallucinate" and "invent" facts (and present them as such). And what does "innovate" exactly mean anyway? You can still argue that the copyright holders never consented to that kind of use of their works. But the whole "innovate" line of argumentation just opens a can of philosophical worms IMO.
I too am an artist and agree with you, but I'd go one step further and suggest that AI is not a tool when it comes to artwork, it does the work for you by following your prompts. For example, a broom is a tool, you use it to clean your floor. Your house cleaner is not a tool, the cleaner does the work, following your prompts. Anyway, that's how I see AI and I've heard the, "it's just another tool, like a pencil or ruler," argument plenty of times and I don't think it holds water.
You make a good point, but I would argue that it is not plagiarism/copyright infringement because the ai takes the techniques and concepts, just like an art student, and applies them. It's not like you can charge an art student for looking at art on the internet.
@@andrewtramel4390 The art on the internet (theoretically) is there either because it has been commissioned and paid for by someone, or it is being used by the artist to promote their services- in other words to generate work for themselves. It's not put there for free, and the fact that it is available for free doesn't imply that it is there to be taken and used by other people in their commercial work for free.
Of course someone can easily take the work and they can learn from it, but they can't (in theory) use it commercially. Even if a student, or let's say 100,000 students use the work to learn from, it is unlikely they will be any professional threat to the original artist whose reputation improves with the more humans who look at, study, share and emulate that artist's work. This leads to more work for the initial artist.
Over the last decade, artists have worked very hard to build these kinds of audiences on social media for the exact reason that it benefits their career in numerous ways. The price the artist pays for that audience is giving their work away 'for free'.
This is not the same with AI. When AI ingests an artist's work, the people who created that AI system are benefiting commercially from the use of that art. The ingested artwork increases the value of their product, usually with no compensation paid to the artist and no consent.
The artist is not credited for their work. The artist gains no reputational benefits for the use of their work. The artist gets absolutely nothing.
When a human artist learns from another human artist, the student can still only work human hours. There is a human limit to the amount of competition that the student represents. When an AI ingests an artist's work, it can pump out images that replicate the original artist's style day and night, at an inhuman rate.
Meanwhile, that AI generated imagery is being used commercially by entities that formerly would have hired that original artist, but now they can get something very similar to that artists work and use it commercially and pay nothing, or very close to nothing for it.
You can split hairs about whether this is plagiarism or not, but the end result is the same: the artist's work is being used without their knowledge or consent, it is being used without any reputational benefit to the artist, it is putting the artist out of work and people other than the artist are making money from the process.
Whatever you want to call it, it's as bad as plagiarism or worse.
I'm so glad veiled sarcasm didn't die out in '17--but it was a close call...close call indeed ;-).
I think since '17,' it's been needed more than ever!
I was going to point out that it was far from veiled sarcasm, since it was out there shouting “look at me!”
And then I realized that I overlooked sarcasm…and my GenX soul is now punishing me for this transgression by forcing me to watch RUclips kids with my little.
Veiled? 😂
This is a natural step. AI is a tool that we should use more. A lot of people seems to think that AI is used as an auto-pilot, when it should be regarded as a co-pilot. Good used is to summarize, filing, and fact checking. I would embrace it to the fullest and is a little surprised that they didn't do it earlier.
This is one dank thumbnail.
Dank how? It's bad?
@@DUNGEONCRAFT1 Dank is good Professor!
Lich's fingernails must be pretty rough looking 😕
@@DUNGEONCRAFT1 dank is a good thing with the kids these days.
Really? I had to ask my son. He said dank started getting used when “sick” went out of fashion. I’m so confused…
I loved the Historical note. :-)
If RUclips allowed pictures, I'd paste in a picture of Data holding a cigar.
AI has already replaced most low level writers who were doing articles at $5 each. I worked as a managing editor with a 100+ writing team focused on high quality work. We could tell the difference when a writer tried to use AI (it just didn't meet our quality standards and we had very high standards).
BUT the reality is AI creates better quality writing than 95% of writers. If you're a writer you really need to work on your skills so you're in the 5%. At that level AI doesn't come close to competing. I suspect the same is true in most other creative fields where AI is being used.
I have found AI useful for brainstorming ideas. It almost never comes up with an idea I use but it gives you a whole pile of different ideas that get your imagination working so you can come up with something original and out of the box yourself.
Me, speaking to the A.I. GM: "I find your lack of imagination disturbing." You are correct when you say A.I. is just a tool, and I would venture to say you, like I, would not look to a tool for inspiration. I shall be the inspiration, and the tool shall assist me in bringing it to fruition; not the other way around as seems to be the trend in movies these days.
I regret that I have but one upvote to give for this comment.
I'm the opposite; I'm terrible at the inspiration part. I can take a bad module and make it good, but I can't take no module and create anything. I can use generative AI to build bad modules that I edit into good modules.
To answer your question: a LMM stands for 'Large Multimodal Model' - it basicaly means a model that can deal with different types of data such as text, audio, video, images etc all at once. to contrast ´chat gtp for example is based on a Large Language Model that only deals with text.
I'm sure the D&D community will respond to this in a calm, measured, reasonable manner
I get your arguments, but I still think they want to make AI DMs. As far as Hasbro would be concerned they wouldn't need to be good, just good enough.
Anybody who ever watched Star Trek the Next Generation will remember what would happen whenever Data would try to tell a joke, that's how I see AI's attempt at humor going.
I'm willing to bet there will not only be AI DM's but AI Players for DnD Beyond watch.
Would make for an......interesting....let's play
AI npcs sounds pretty awesome actually
Would love having AI players since I hate humans. Finally machines of loving grace.
NGL, I'm morbidly curious to see how AI would handle a failed sanity check from a heavy Far Realms encounter. Would the madness spread? "Welp, there goes humanity!"
Why stop there? We can add AI watchers too and cut out the human component entirely from entertainment. Think of the efficiency!
Question: If they say now that it's 'only' for video games, what keeps them from going back on that?
It is true that the Industrial Age created new jobs to replace those jobs lost by the transition from the Agrarian Age, but those jobs were detrimental to health, created overcrowded cities, destitution, hunger and a raft of other issues such as alcoholism and petty crime. When people lived and worked in the agrarian world, their jobs were healthier, and food was plentiful.
Imagine for a moment if someone came to your home and took all the clothes out of your closet. They tossed those clothes into a wood chipper. They picked up the shreds and dipped them into fabric glue and twisted them round into giant wads that looked vaguely like shirts and pants, and they charged you for these "new" clothes.
That's the current model of AI. I don't even like using the term nexus at some point, we ARE going to create real AI, and we'll have to call it something stupid and obtuse because the term Artificial Intelligence will be tainted and politicized.
Yeah, AI has been in our tools for a couple decades by now. Photoshop, Maya, game engines, word processors, etc.
I think most people are objecting to generative AI, largely because it's doing something AI hasn't really done in the past.
LMM: Large Multimodal Model. It's a multipurpose AI. It can handle everything, text, pictures, sounds, and video. An example is Google Gemini. This is opposed to an LLM (Large Language Model) which deals only with understanding and generating human language (ChatGPT, for example).
"There's a stat block. you can KILL ME." LMAO Professor DM you had me right here and I'm going to own this.
Why pay workers when you can enslave A.I.
That title /slow clap
Hahahaha bonus points for explaining what a joke is.
Thanks for watching!
word!
Brilliant!
WOTC: "Pay no attention to the man behind the curtain..."
There is no man ...
Thank you for your balanced take on the subject. The doom sayers for A.I are really putting a damper on what could be a fun cool tool if they just embraced it with caution.
Yeah. I have faith it will improve our lives. Also my retirement. I own stock in NVDIA.
I think that like everything else, DM's will ultimately utilize the tools that interest them & supplement their particular style of play.
When I first played D&D in the late 70's, my DM's were "typical nerds" (intelligent, creative, socially awkward) and older than us. As they graduated high school or moved on, I, being the 'next-nerdiest' of my friends, became the full-time DM running the classic modules. Soon I was stringing these together in increasingly complex story arcs and substituting more and more content--what we now call homebrew.
To your excellent point, Professor DM, this was an adaptive response to my player's input: our collective style.
I do not believe that AI can replicate that specific type of synergy--at least not yet.
As any DM worth his dice will tell you, go ahead and "borrow" (and adapt) any tool out there to achieve the results that best meet your table's desires.
Keep on rollin'!
In my Chat GPT experiences (limited though they have been), I've found three things to be generalizable statements:
1) Asking for factuality demonstrates the creativity of the engine, because it outright makes things up (including references to non-existent articles/books!) but they *look/sound right* to the untrained/uncritical eye/ear.
2) Asking for narrative, though, is pretty awful, and it can get very nitpicky on particular characters and attested lore/sources (it would be good at the quiz show "Um, Actually"!), but then asking for poetry also generates doggerel that may rhyme fairly well, but is boring and cliche.
3) Something that it does quite well, though, and can save some time and agonizing on for people like me--who love to write and do it well, but dread writing certain things--is cover letters and short business e-mails. They can end up being a bit jargon-y and use a lot of buzzwords (which, unfortunately, HR people and business types love!), and of course they need to be read over and checked and certain small things generally need to be altered or expanded, but this is a huge time-saver.
Thus, I suspect it might be able to describe certain things that a DM might find useful if they're not sure what to say about certain rooms and such, but creating a cohesive narrative is probably beyond it at this stage.
AI image generation, though, is wonderful as far as it goes, especially for those of us who aren't great with art, and if one is only using it privately and not in a publication, I don't think there's anything at all wrong with using it; that being understood, though, not all programs/websites/apps in that regard are equal, and some are far better than others (including amongst the ones available for free), and the results on these can be surprisingly good and creative, while also being occasionally disastrously awful but good fodder for laughs (even independent of the tendencies to have super/subnumerary fingers and toes, or even whole limbs, etc.). I recommend one called DezGo if you're interested in experimenting with it (dezgo.com); it can do Elves and humans/humanoids well; anything else is a bit of a crapshoot. ;)
Having played around with Artbots and knowing how to draw myself, trust me, you're better off spending the time learning how to draw. There's a lot of videos here on RUclips about getting started. I'd go with Marc Brunet's video since he goes step by step on how to do it. Search "marc brunet 30 days" and it should pop up as the first video.
Are you, perhaps, suggesting there is a SPIRIT to the art of role-playing that a machine cannot possibly replicate? Such materialist language, I wonder to what degree intentional. Great story. Love your coverage.
Hahahaha the 'Definition of A Joke' was a master class moment! Had to relisten to it, it was so good 😂 The intense level of dryness in the delivery*chef's kiss*
Thanks, Prof. Great thumbnail.
Remember when artists in the 80s and 90s were big names. Frazetta, Elmore, etc. I mean Image comics artists even had commercials on TV for levi buttonfly jeans!
Same with Hollywood, I dont think the names carry so much wait as they used to with say Bruce Willis, Sigourney Weaver, Linda Hamilton, Arnold Schwarzenegger.
I think it is related. I am not sure what to call this phenomenon but something is definitely different. It used to be the stars that pulled us, but now that everything is a sequel, prequel or remake, we just follow a IP instead of a artist
the suits have learnt that they can't own the name of an artist as well as they can own a franchise
Also the sheer nature of the world we live in makes these identities very VERY culturally isolated. It's not that we don't have "movie stars" anymore... It's just that there are smaller and smaller pockets of micro stars. We don't ALL watch TV shows at the same time on limited available broadcast schedules. and if we wanted to watch the new movie we ALL went to the theater... Now all of that and more is at a finger tips. People are just more isolated cultural in some ways. And yah some of that is by design of the corps.
I use grok for hooks from specific settings prompts, monsters specific to areas, spellbook generating for wizards and dragons that are based on a theme for the wizard or dragon.
Thanks for the fair and measured take as always, Proff
I love all Dungeon Craft videos!
In other words, Robo-DMs.
I think this is inevitable. It doesn't matter whether humans are better. There aren't enough of them. For the VTT to succeed at volume, they probably need robo-DMs. The unpaid human DM is the weak link here. Also, I'm sure we all agree that fully detailed 3D VTT maps are too expensive to be created on the fly. A robo-DM could do that. This will prevail for economic reasons -- it is way cheaper; no one wants to pay for the alternative -- so for it to exist, the robo-DM will be a thing.
Dungeon Craft is my FAVORITE RUclips Channel👍👍👍.
That's kind of you to say. Because there are a LOT of good ones. Made my day.
I fed the descriptions of 20 rooms into AI and asked it to assign a difficulty factor for each room (you would call it a room dc ) and in 2 seconds it gave me both the DF and an explanation as to why the room was assigned the DF. This saved me an hour or two of my time that I could use to create game mechanics I needed for the special encounters. I love AI as a tool. They Just announced AI chips (like physics engine chips) to go into new PC going forward. For Video games it will soon be used to generate areas that the player wanders into but the DEV did not have time to make. "I for one welcome our AI overlords".
Alright Mr Gen X. You're wearing ripped jeans underneath that table, aren't ya.
One thing doesn't pass the smell test, Professor; "video games, down the road" is not a real job position. Everything on the posting suggested utilizing and "road-mapping" AI sooner, not later.
Hi Professor, excellent video. LLM stands for large language model, aka the type of AI algorithm that Chat-GPT and stuff like Midjourney are based off. It's a technical term, not out of place in a job posting.
This is in order to differentiate it from other AI models (from data science prediction algorithms, to Chess players, to general AI).
Thanks for sharing!
The text said LMMs.
@@TheLoneQuester large multimodal model
AI is veeery useful for creating Minor npcs and to do research for creating random Encounter Tables etc. I think the first step would be GM tools for the vtt
I work in the video game industry, we've started using AI tools to help with game development and QA, I've had a chance to work with the tools and the teams indirectly and I can confirm the description of what they're asking for sounds a lot like the job description of the engineers I've worked with, it sounds like Hasbro's game development interests are just trying to catch up with what the big video game studios are already doing.
LLMS in the job listing most likely means "Large Language Model Systems" which is the heart of most modern AI programs.
Wizards has been talking about an AI dungeon master since 4th edition and the precursor to dnd beyond. It will never be a replacement experience, but may be able to faithfully run groups through premade modules (perhaps stilted, without a special DM flair). I think this is fine. No replacement for the real thing, but for example I would love a tool that runs me through some old adventures my group has no interest in playing.
I have no objection to videogames based on an RPG, for example the Pathfinder ones are great. It's a different genre of game though in my opinion.
Jokes died in 2017… the same year your oldest video still on RUclips was published…
Just kidding! I’m a huge fan of the channel, and not just your drama videos! The campaign recaps, XDM recommendation, and class reworks have all been super inspiring for my game. Thanks a bunch for your videos!
Great coverage on current events Professor!!
I like how being able to speak about how using AI is ethical is part of the job description as one of the elements behind being hired to work. I think it's because there is so much fan backlash against AI generated content that professionals have to justify their jobs to the fan community.
This is disappointing because in my opinion the fans need to calm down about AI generated content. What matters is not whether a few precious jobs are saved (protectionism just doesn't work) but rather whether the content being produced is any good. We should be grateful that AI will allow more quantity if not quality of content being produced so that we can have more to consume.
I agree with this making sense for the VTT. I could see this possibly being for tables that want to play D&D but none of the players want to DM. Based on playing around with current language models, I agree that the tech isn't ready to take on the seat of DM, and I'm not sure if it will be any time soon. I could see it working as a way to make NPCs more interactive in the VTT.
Oh, only for video games…my industry. Thanks WotC
Years ago I was in a sculpting competition and was up against the digital printing stuff. It felt like going to a weight lifting competition against a fork lift. The digital artists mocked us with "Art is art, it does not matter were it comes from!" Now that they are in the same position against AI, it gives me a chuckle, but I'm still not a fan of AI.
The environmental issues with AI may be what destroys us before our robot overlords can.
As someone who is rather fond of AI and use it for hobby stuff and messes with it as a hobby ‘games’ using AI as a major part of the experience mostly as a co-writer (imagine a TTRPG in which the players and the DM share the same jobs) of it already exist. AI Dungeon was kind of the front runner of that with effectively advertising itself as infinite text adventures through a LLM with a few others coming up over the years.
These however to this day still have more than a few hang ups. Namely they have got to be handheld or they go off into weird tangents/break from logic and if you don’t shackle the AI pretty heavily they tend to get really naughty in to what they will generate. But censuring the models usually just makes them bad at everything as even perfectly mundane inputs cause it to freak out and stop.
I wouldn’t trust AI to act as a DM on its own unless you are up for a particularly trippy experience as dead NPCs are just casually back, the DM constantly telling you that you randomly die, and a adventure that will regularly try and devolve into smutt at its nearest convince.
2:20 maybe that's a typo for LLM (large language model)?
Large Multimodal Model
I'm guessing LMMs was a typo of LLMs, i.e. Large Language Models.
I still very much love all Dungeon Craft videos!
ChatGPT Prompt: Write an original joke about a RUclipsr with a channel named Dungeon Craft and make sure the joke isn't just a derivative of obvious and well known jokes.
ChatGPT response:
Why did the Dungeon Craft RUclipsr refuse to play Monopoly?
Because he said if he wanted to spend hours arguing over imaginary property, he'd just read the comments section on his latest dungeon build video!
I don't know... AI seems to be on par with the creativity of most RUclipsrs 😂
So, right now we see D&D as being a social experience where we're coming together to hang out and have fun with friends, playing a TTRPG. The experience is communal and flexible, with just enough rules to identify the boundaries and stakes, creating a sweet spot where the magic happens. That version of D&D as we know it, I feel, is safe from AI. I don't think you can replace that. Spontaneity, personality, expectations... an AI could be 'fair' and do what we want, but I feel it would create a more video game experience, since the AI will always have less agency than the live player. I don't think many players would want a table where an AI has more agency than they do. But we can trust, know, and grant that position to a live person. And once AI can host immersive experiences, I suspect we'll see many of the other trappings and boons of video games and even VR coming along with it.
Considering that, I think AI will create a new space for TTRPG that exists between live D&D and video games. Single-player adventures with increased player agency and freedom, or very highly-trained modules. Even if players have to accept that if they go too far outside of the boundaries of its training data, it may have to tell them no, the same way a live DM might tell a player no to something they have no prep for, if it creates an issue for them running the campaign. That may well be an acceptable tradeoff for a currently unrepresented demographic of potential players who have difficult or limited schedules or other accessibility issues that create a barrier for playing live D&D.
AI-run TTRPG could also be a great learning tool for new players to test the waters, or try out new systems. Imagine being able to go through a solo adventure, being given advice and guidance tailored to your play and knowledge. Getting to familiarize yourself on your own schedule and time, doing whatever you want, could give players the confidence and readiness to hit the game running when they have limited time to play live games with friends.
I'm fully prepared for companies like Hasbro to make horrible products and try to use them to reduce overhead and charge players extra for less. "We made this shoddy campaign, but you get access to an AI companion that will patch it however you like! Probably!" But there's a lot of amateur innovation with AI that's surprisingly accessible, so I fully expect that fan projects to make a GOOD AI DM will produce compelling results before long. And as a more approachable expectation, will also create AI companions for DMs that can help them with campaigns, generating maps quickly, creating custom art, writing songs and speeches and descriptions for otherwise-rote stuff or patching in to libraries of artist-created assets to customize them for your use.
We face a lot of ethical challenges with AI, and companies will try to bend it to both cynical and immoral spaces. But I honestly believe in the community and creative power of the TTRPG crowds to get involved, stay vocal, and innovate uses that will support the hobby instead of damaging it.
I'm known of the Twitch channels i hang out in, for Dad Jokes and Puns! : )
I once sent a list of my top 10 favorite puns to 10 of my favorite friends, and asked them to vote on their favorite pun... after tallying all the votes, to my dismay, no pun in 10 did
I truly believe most companies are jumping on the bandwagon, within a few years it will be everywhere.
I just finished my review for the most recent module today, and I'm glad I'm not completely out of pocket with my opinion on WoTC as of late...
Man, that explanation of what a joke is really got to me. Brilliant.
They have given up on lying about AI in their products. They’ve likely been using AI to write adventure paths and other source books as well.
I have long since stopped supporting WotC (since 2017) and I won’t purchase anything labeled “For Use with 5E” or similar.
Let WizBro’s TTRPG die on the vine. Support OSR & Indy that has divorced itself from 5E.
I don't think so, but cannot say for sure.
@@DUNGEONCRAFT1 give them time, they’ve already said the VTT will have an AI DM, and the VTT is the future of WotC D&D.
I like Chat GPT for generating game ideas, and exploring a few different NPC concepts. If I know I want part of a game set in a swamp, but I'm not sure what I want for the challenge/bad guy/adventure hook, it's very much like a random adventure roll. That said, considering the amount of fresh water it takes to operate the machinery, I'm not sure it's worth it. It also has canned phrases that crop up over & over, so anyone using it a lot is going to end up with that voice permeating their own adventures. And I think I've exhausted it's store of riddles (which are not very good).
Thanks for sharing!
I enjoy all of Dungeon Craft videos
The funniest comment I read about AI: "I now believe in the concept of a human soul, now that I've seen art made without it."
The advent of photography is a really interesting technology with parallels to modern day AI.
Originally it began relegated as a research tool, before the pictorialist movement elevated it to an accepted art form.
People feared it would replace painting, claimed that it would ruin livelihoods, and insisted photography couldn't possibly be art. The reality is photography changed things, sure. The world adapted to the new accessibility photography brought - but life carried on and progressed.
These tools both changed the accessibility of image generation, and from an economic viewpoint you can't increase something's abundance without decreasing it's value. In my experience, many people who dislike AI inherently dislike how this ease of access implies a devaluation of existing works, even when they find it hard to put into words. However, people adapt and end uses change to fill niches.
Photography has since developed to a level nobody back then might have dreamt of, and AI may well develop on a similar trajectory in the future. However, the box is open and all we can do now is adapt.
I'm always wary of saying AI could NEVER though, rather than AI currently can't!
You are right about keeping all the AI machines running. They showed a relaxing future of it in the Matrix. Chaplin's 1936 movie showed the bright future brought about by the industrial revolution also. Luckily the end of the agrarian age didn't do horrible things like raise pollution or excellorated climate change.
Nope, all smooth sailing on the AI boat.
Dungeoncraft is awesome! Thank you Professor for your insights. Very helpful
Thanks for bringing attention to this issue. I hope AI gets better as a tool, but I really struggle with the ethics of how it’s being used currently.
AI is unlikely to get better as a tool. It might have already peaked. The problem is diminishing returns, and the bottleneck of training data. All the training data used in image generators like Dall-E and Midjourney were obtained by scraping sites like Art Station and Twitter. That was the low-hanging fruit, though. First off, there is a cultural bias in the data samples and in the people collecting the samples, and it is reflected in the style of images AI can generate; diversifying the training data is going to take more legwork. Second, artists don't like their work being used in training data without their permission, and they will do everything they can to prevent it. One tactic is using programs like Nightshade to poison AI's with patterns hidden within your artwork's pixels. Third, the proliferation of AI generated images on the internet means that the potential pool of training data is now contaminated. The kind of web scraping used before will result in AI's being trained on AI generated images; this may create a feedback loop, or reinforce the messed up hands problem.
All in all, AI generated images are only suitable for cheap slop. I would expect as much from WotC. I will continue to boycott their products, and I will continue to support my favorite artists by buying their books.
I understand why professional artists are worried about AI, but my experiences at the table have been very positive. My GM loaded his preferred random tables into an AI, and it is able to instantaneously produce a very complete profile of an NPC or lacation , including an image in a second. It's not publishable quality art, but it is more than sufficient for play.
I like this channel more when it focuses on the hobby itself. Really not thrilled with how much kickstarter promotion and hasbro news ProfDM is covering lately.
need AI to create a stat block for Luddites
I just enjoy writing stuff and AI feels like I’d be cheating myself of something I enjoy. But I feel that way about something like The Fantasy Adventure Builder or pre-generated outlines, too.
I work with AI daily, and it has been advancing rapidly. Currently my team is focused on observability, validity, and proper Prompting techniques. We have already replaced a few traditional ML models with 3.5-turbo, and 4o is going to be replacing a yolov5 model we use. Basically I can see the technology replacing even more then can today in just a few years.
As a DM it still has trouble building dungeons, quests, and adventures, however I have trained a model using my campaign's notes and its become my perfect Co-DM. Whenever I can't remember a detail or quest line from years ago it can find it for me.
Thanks!
Here is an original joke from ChatGPT: “Sure, here's an original joke for you: Why did the scarecrow become a successful motivational speaker?Because he was outstanding in his field!”
To me, the excitement of AI in dungeons and dragons is how it can be used with the NPC‘s. I envision games on a digital level, becoming more and more self supporting, and the players will enter into these environments. For example, let’s say a village is created with a bunch of NPC‘s, and the AI is going to be working the NPC‘s. So this village would be almost like a living village or a village with real people and when the players come in, they’re going to be able to interact with these people on a more detailed level. It’s almost like a sandbox environmentthat heals itself up and keeps itself going, and players are visitors.
Thank you for the thoughtful, balanced analysis! I think there's a bit of a moral panic about AI in some corners of the TTRPG hobby. I tend to think that AI will be harnessed to augment human creativity rather than replace it. The rumors of human DM extinction are greatly exagerated!
AI will probably be able to replace humans in some jobs, but it’s got a long way to go before it can do that. I think we will probably start seeing it in our lifetime, but it will probably be future generations who will really improve it.
They also had an artist steal creative work material directly from other artist's work, like from Cyberpunk works, for MTG.
AI DMs would be like open world video games like the Witcher 3 or Elden Ring. To be able to adlib on the fly outside of your basic video game limitations is still well into the future, at least enough to satisfy gamers.
The technology advances exponentially. It's right around the corner...
until the program actually understands what it is saying or being told, it'll be very jank and ultimately worse than a human GM
that's on top of removing human interaction from a tabletop game that is effectively built off of human interaction, which isn't a very smart idea since it will effectively gut the core of what ttrpgs are about in the first place.
what makes open world RPGs feel good is that everything you find has been placed by someone, every funny or interesting dialogue has been written by someone
Generative AI is incredibly dull and uninteresting
This is the audience they wanted, and now it's the audience they have. Broke, woke, and angry about everything. Good luck with that.
I've found AI very useful for generating a dozen ideas for something. Usually most of the ideas are just bad or fail to take into account the circumstances of the situation. Sometimes there are no good ideas in what the AI produces, but even in those cases, it will almost always inspire a good idea in my head.
Pretty much all of my players are using AI to generate portraits of their characters, and I've been using it to create portraits of NPCs and landscapes to help my online players see what I'm talking about. It's not great, but it is affordable, where my own "staff" artist wouldn't be. And it's pretty good at taking a sketch I've made and taking it 90% of the way to finished, where I can do the final touch-ups with GIMP.
I don't understand why people talk about AI game masters as if that would be the only way of using AI in ttRPGs. Surely to make a fully independent GM is not quite possible with the current tech (we will get there one day though), but AI tools could be a great help for GMs while running the game. For example an AI could listens the game on the background. Then GM could any time ask help from it. For example it could ask AI to invent an interesting twist to the story or to fast create a suitable NPC for a certain situations. Or AI could even totally take over some background events: what are the 10 most central NPCs or factions doing while the PC are busy with their things, and how that could become visible to the PCs.