AI News: The AI Arms Race is Getting Insane!
HTML-код
- Опубликовано: 14 апр 2024
- Learn AI With Me:
www.skool.com/natural20/about
Join my community and classroom to learn AI and get ready for the new world.
🎶My AI Music Playlist:
• Best of AI Music
#ai #openai #llm
BUSINESS, MEDIA & SPONSORSHIPS:
Wes Roth Business @ Gmail . com
wesrothbusiness@gmail.com
Just shoot me an email to the above address.
Seems like everyones main use case is... use my ai to read other ai generated content to then generate ai generated content!
Most human work is derivative, I guess there's no difference here, just speeding up the churn.
@@MrClubSandwich a system that is not real-time adaptive would always ALWAYS BE A TOY...SINCE IT CAN'T SURVIVE IN THE WORLD EVEN ON THE WEB-LIKE INTENSIVE CYBERWAR-FARE, OBSERVING AND CONSTANTLY ADAPTING TO STOCK MARKETS AND ISSUEING SUCCESIVE TRADES ALSO IT should be capable to SUGEST AN EFFECT PROTEIN-SERUM FOR MOLECULAR BIO-ENGINEERING AUGMENTATIONS.
AND DESIGN 3D a functioning HOVER-BOARD...
Ouroboros.
Yes, I think generative AI has plateaued. It can’t reason that much, although it’s good at ideas generation and content generation and finding some correlations. People are trying to milk the current AI for more than is possible, and to be honest it’s becoming embarrassing.
@@mikezoopergotta teach it pain so it has a reason to get creative
9:03 100% on the ADHD comment. Please build that. Basically an executive assistant that monitors what you do, sees, hears, has access to all emails and correspondence, and can know your schedule, devise priority lists, etc.
The whole reason I want agents 😂
I dearly need an existance admin agent.
and prompt you if you deviate too far off task
This is the reason I've been researching agents lately haha. Adhd peeps unite! 😂
AGI: why the hell should I spend cycles keeping this misfiring monkey in check, I can do everything it does more efficiently. Fuck it, launch SkyNet!
If a HUMAN can read or do something online, an AI should be able to do the same. ❗❗❗
100%!! Plus artist use other styles all the time, not complete rip offs, AI learns styles, it uses these styles to make NEW art. How is that any different the when I look at art and love certain styles and recreate that myself? I love certain artist and 100% my art can lean in that direction. How can you stop anyone or an AI from the freedom of looking at something? or watching something?
You can make things from other styles to make your own, and for private personal use you can paint a picture of Spider-Man or Mario, Pokemon, whatever, it's called fair use. Of course if you recreate something copyrighted in full and try to sell it or distribute it, then that's against copyright. But I can draw the Joker all over my house, I can't make a comic and profit from it with the Joker as my character, or make a TV show with Spider-Man or sell a T-Shirt or a mug. They are fighting fair use, not distribution and sale because they fear it being out of control.
Artist always steal from past creators! In comics you can easily see what styles artist stole from, sometimes it's almost identical in style.
How many new wave artist stole from Kraftwerk? Michael Jackson stole the moonwalk and made it better, The Flintstones was stealing from the Honeymooners... over and over... bad artists copy. great artists steal. Go after copy and profit, not watch and learn.
I can watch ALL the RUclips I want, if they are scraping data, then yeah, that's wrong, maybe that's what they are doming?...but just watching? storing a tiny vector of an image, not even the image itself? give me a break, let the AI learn everything or the entire thing is broken, just fight the people that try to sell copyrighted content in FULL, and leave fan fiction and fair use alone too.
Please add links to every topic you are talking about, that's a very important service to your audience
"they have a little safety harness .. so the robot can't escape" 😂
Hahahahahah......ha👀
Thanks for the consistency and hard work Wes! Love getting your videos so regularly.
Thanks for the mention, Wes! Would you mind if we shared a short clip of your video on our socials? Credit given, of course. Either way, thanks for checking us out and let us know if you have any questions!
sure thing, go for it!
I might do a deep dive into the product later.
@@WesRoth1hr interview would be nice
It's fascinating to see how AI isn't just about speed but also about smarter, more efficient workflows.
how is that even remotely fascinating
@@Proprogrammer001 My AI wrote that comment. I joined with them as soon as they arrived. I suggest you do the same
Oh! I am actually working on that exact thing, a system built specifically to help people with ADHD to achieve their goals! I just have ADHD myself so progress has been very slow.
Heh, ironic.
I have a feeling a lot of ADHDers around the world are working on the same thing right now, with the same problem standing in their way to get it done. 😂 I imagine mine like an AI that is basically doing the stuff parents do for their teenagers, just for grown ups. I just need someone to remind me of the stuff I wanted to focus on, and if I'm not doing it, have a conversation with me about the reasons or stuff that could motivate me.
@@missoats8731 Would that work for you? XD I would just probably dismiss it XD "Later"... "later" XD
@@gweneth5958 That's where the conversation would come in. Maybe it could lock my screen until I a) give a good reason why I'm not doing what I'm supposed to or b) start doing it :) If nothing works, maybe it should be able to punish me or something like that. But I guess it can't be perfect and can only do so much.
@@missoats8731 😂 "And the phone was locked forever". I know what you mean. There were apps like that to make me keep writing and the phone didn't work while that time... well... you may guess what happened. I guess, as long as you really want to do things, then it will help. I think, it would best if it would talk to you through voice than with text then. At least, I think, that voice would work much better with me than reading it, like a voice suddenly saying "Are you working on xy?" and if not then the reasoning starts =D
Timely and focused presentation. Very informative. Thanks 🙏 Wes.
It's not the end of artists, just the end of money driven artists
Art for everyone!
That is my thought. I guess that some of them just don't have a clue and just follow the crowd, like people following influencers saying stupid stuff and they keep repeating it...
Thank you for the long vid Wes
Read the books first, it’s part of a trilogy, some of us feel is the greatest sci fi ever written. First book; 3 body Problem, dark forest, death’s end. From a theoretical physics point of view fuc… amazing.real people real world.
Agreed, agreed, agreed.
Love that you’ve added your face to vids lately so ppl know the voice isn’t just AI generated lmao. Says a lot about how well the space is developing
Right at the time when it is getting possible AI generate a talking head avatar for a voice 😉
Shockingly efficient. Thanks Wes
I like your take on using this as a filter. We're quickly approaching a lot of paper in the wind when it comes to automated articles and spam. Nice take.
As someone with ADHD I have also thought how invaluable an assistant AI would be to help with reminders and task focus.
I understand why something like that seems good but it isn’t in this case. You’re seeking something outside of yourself to mend a problem that stems from you. If you’re the source, it doesn’t matter how good the tool is, YOU still need to do the focusing. If the tool helps you, it’s YOU who made the change and not the tool. Therefore you wouldn’t need the tool. Sure it’s meant to help but hopefully you see my point.
@@thephilosopher7173 by that logic you should dispose of your shoes, your clothes, your toothbrush , your shampoo, your computer, your car, any insulin or antibiotics you might need, etc... Unless your point is to suggest that ADD is mere mental weakness. Is that your point?
@@MattHabermehl You could but certain cases that wouldn’t be practical. My question is: how are you going to learn to focus, by having an AI tell YOU to focus, when YOU have to do the focusing? If you have to do the focusing and the AI helped you, then all it did was give you motivation to make that mental effort. Sure if you feel it could help then use it, but my point is that it was all up to you in the first place.
Seeking this as a tool to depend on for the rest of your life will basically put your autonomy for attention and focus in the hands of AI as if it’s AI doing the focusing and not you. Crutches help you rehabilitate your legs but once you walk again do you still need them? Was it the crutches that repaired the muscles or was it your body? Is it the Ai helping you stay on task or are YOU staying on task?
@@thephilosopher7173 I don't know if you know this but people with ADHD lack executive control function in their prefrontal cortex. It's like telling someone who's disabled that they'll never learn to walk by relying on a wheelchair. I'm not saying that people with ADHD can't learn how to improve AT ALL, the same way people without ADHD can; but our capability in this area will always be crippled.
I make regular use of "hey Google" to set tasks, reminders and add things to my shopping list because if I don't, things get forgotten, and I don't function - my whole life doesn't function as well. I've learned not to cross something off my shopping list while walking from aisle 4 to aisle 7 to get it, because experience has taught me that I might get distracted on the way and forget all about it. I'm not going to stop using a tool in hopes that maybe I won't need to rely on it as much, one day off in the distant future, severely hamstringing myself in the mean time.
I get what you're saying and why you're saying it, but unfortunately what you're asking is just not based on a realistic understanding of what ADHD is and its limitations.
@@thephilosopher7173 lol spoken like someone without ADHD. It is an attentional impediment, for sure, but not correctly characterized as simply a lack of focus, as if additional effort when trying to focus would fix it. It shows itself in adults mostly in moving between tasks. It manifests, from an outside observer's perspective, in carelessness about details or a kind of forgetfulness regarding details when performing one task and then moving to another task in haste, or if distracted by something perceived to be more important or urgent. Certain strategies can be used to mitigate the effects, but they are all technologies of a sort, whether it be a memorized mental checklist, or a written one, etc.The desire for an AI version is just a desire for more efficient tech.
I like your take on the personal filter use case
Amazing, thank you.
LOL Wes that was brilliant!
I have deaf friends online that could benefit from a tool like this. Id imagine them being able to understand videos without captions better using this.
9:00 Man... I feel you. I'm working on exactly that because I need it too XD. Flat out, I'm all in on the home JARVIS/HAL system that's watching me for visual/audio cues to log/assist. If it sees me walk in and out of the same room 3 times, I want it to ask what I'm looking for so I don't forget on the 4th pass without me having to remember that it's even watching. I'm one of those use cases. To hell with privacy, let me have a productive vibrant life.
I don't know if you've ever read the book "Fall (or Dodge in hell)" by Neal Stephenson, but during the first third of the book, or so, some of the main characters had this functionality built into their Augmented Reality glasses. They never called it AI, but it was a customized and curated algorithm that would filter out any information that the user didn't want and would notify them of information they did want.
Love the idea of a filter VS of a pump!
lol the robot duck. I’m so old. I want to see an AI trained on just the roadrunner/coyote content!
'3 body problem' - better read the books first, they are worth every minute...
@ 9:40 I have done that multiple times, on the bright side, "matter and energy is conserved", even though it was not recorded.
I teach online and record videos for my classes. Once I was having a bad allergies day and had to stop recoding every now and then to sneeze.
I missed pressing the record button after one of those, and ended recording for two hours... With the result of a video that contained a compilation of all of my sneezes. I got so pissed that I deleted that video. I now regret it. I have students who would pay to see that 😂
Thank'e!
Wonder what Isaac Asimov would think if he learned the “era of AI” just means you’re on 30 waitlists.
Love your videos. I learn something new every time.
In this video, I was frustrated with trying to read along with you, your face image was covering a lot of the screen views.
Omg wes about the adhd agent, I agree 100%. Someone please make this.
I’m so glad I live a simple life as a truck driver. I love listening your videos to discern how things are going in general and am very thankful I’ll never utilize these agents.
That was weird, my phone screen went all matrix green tinted while i wasn't looking for some reason. Thought wes was doing a matrix agents reference or something until I turned it and saw the comments in green too 😅
Thank you Wes we truly appreciate you very much.
Stay the course.
Whenever you may be “in doubt”
just do whatever is in the best interest of others.
Especially if it comes at a personal cost.
There is definitely a race… a very desperate race straight to the bottom.
Do not get tangled up in that one.
Take care
Jeremy :)
13:29 I find amusing how the robot shows a gesture of joy cause he managed the task
3 Body Problem - I implore you and anyone reading this to READ THE BOOKS first. and it’s a series. And there’s a Chinese series too 😊
@9:35. Facts, I definitely could use a ChatADD
STUNNING
Defense department came without a password. I even used customer service provider
I've noticed that we don't have to wait long for the interfaces and models to improve by many times, making it so much easier and faster to generate content and publish it. At this point I'll just wait for Sora and GPT-5 to come out, then I'll be ready to get into content creation. It's only gonna get easier and easier to do it, no point in sticking with one method when it will become obsolete in just a few weeks.
Thanks for the interesting updates. The sound is too low volume.
Japan is actually quite strict with copyright laws. They are also quite open to robots helping alleviate issues arising from demographic shifts due to low birthrates.
You're getting better at each post. Congrats 😀 (great ADHD comment!)
The entire concept of copyright needs to be revamped from the ground up. It has been abused for decades now to enforce monopolies FAR beyond what the original legislation (in the US at least) intended. I wrote a research paper in college 20 years ago about how bad it was then, and it's only gotten worse since then. I wish I still had a copy of that.
Hahaha, talked for a half hour to a dead record, that was funny. Love you Wes.
OMG the ADHD comment got me. Was listening in an Uber and just busted out laughing. Got some looks. Just this morning I wrote a comment in my work’s Slack making fun of myself for going to boil water for coffee three times this morning and then getting back into stuff (forgetting that I boiled water for coffee).
Yea, we definitely will need an "A.I. filter" similar to email spam filters. Hopefully the filter is good enough to completely block I.P. addresses.
Batch API is not for "asynchronous agents". It is to upload prompt lists that asynchronously complete, that in no way shape or form interact with one another...
re copyright law… you can probably think of it as a copy of the original is compressed into the model. How much remains after compression does not matter. It’s in the model.
chills
Custom filter and algorithms would have been easy to do for any company with traditional means already. Looking forward to systems that serve me, instead of just being optimized to bind peoples time (for ad exposure chances)
Sorry, I have to correct and clarify one thing--that "robot duck" is Gizmo, and he was not a robot. He was a bit like Robocop in that he was a cyborg. Sorry, nerds gotta nerd out.
Don't worry, I am always annoyed that people don't get that a Cyborg is still a human being XD They think a cyborg is a humanoid robot.
13:19 knew a guy who worked at a sorting terminal and by his account of how his coworkers would handle packages this bot would blend right in.
"RPC, I don't know what that means" = "i don't know how computer systems that call remote procedures work"
Shocking
Also dreaming of an ADD assistant too
It's being said a lot but I'll say it too. The 3 Body Problem is one of the greatest pieces of sci-fi ever written. The show is REALLY good, but all true lovers of sci-fi should read the books.
Training data is programming. It's a new kind of programming based on transformation, but an honest reflection on what is technically happening here should result in the conclusion that "training data" is no different than pre-compiled code.
You could build an AI agents team to get another pic of Sam Altman
I’m finding perplexity to be pretty good
If you keep focusing on things that makes your "life" simpler, you start to think less, learn less and be wary about outsourcing thinking and gathering information
The humans in Wall-E comes to mind.
Wes, Your the best AI news guy by far. I'm sorry to be that guy who is blowing smoke up places they don't belong, but I've been paying attention and the only information source I find helpful and informative is you, my guy. Good on you and I look forward to seeing your future rise in the AI world.
2x playback will help you and Wes
yes i want my agent to be like that little floating elf in Zelda. lets me know if i am doing incongruous things
Reka Core just spills out random video summary, when asking it to summarize a given youtube link. each refresh (with the same link) hallucinates different content. Not tuned to know its own abilities.
Gizmoduck....GIZMODUCK !
Volume on max, can just hear u if i stay real still
what do you thing about cyborg and neuralink, thanks ;)
Couldn't that 'ai' simply detect the 'netflix' written in the top right corner on that video, then search 'netflix movie number' which would lead it to all the rest of the info about three body problem series including all sorts of descriptions etc.
working on it :) .... it will even talk back :)
The "default option" right now is that people's copyright claims are getting shoulders shrugs. Also aren't there plenty of cases still where you can literally see copyrighted material manifest within model outputs not just variants?
If you are talking about asking an AI facts, there are only a limited number of ways to state the same fact. If one person's copyrighted work happens to state a fact in the most obvious way, then it really shouldn't be allowed to be copyrighted in the first place.
BTW, pronouncing ETH Zurich as “Eth” would be a lot like calling MIT “mit”
I like your cartoons at the front-end, Wes. Wes I don’t know if that’s exactly where this is going because I was talking about creating media devices that generate their own contact and you aren’t under any obligation to look at anything from anyone. We all start a product line and we protect ourselves from the rest of the businesses. AiPod. Think about something about the size of an iPod that you could stick in your pocket but it’s your computer or cell phone and it’s also running your AR glasses. If we’re doing this right, you’re not going to need an interface, but you’ll just tell it what you want is a desired output the best you can whatever it is you need from a research to a television series. If we develop things correctly, AI will also make us smarter and not let us get completely lazy as it’s helping us. It teaches you something every time it hand you the information. Even how AI poses a question or delivers the information will help you develop as a human also making the interaction between both computer and human more seamless. I started thinking about the fact that this could actually change fashion because we could use networking running through our clothing and spaces for battery packs. : ). Part of that will be to have sensors that are touching your body, but you can pull off before you wash them or replace them. I think eventually will get all of it put into the threads and we will be able to make Weavabld electronics that are clothing. It’ll probably be graphean and also act as body armor.
By YouSum
00:00:00 Advancements in AI models and AI agent workflows.
00:00:11 OpenAI introduces batch API for asynchronous tasks.
00:00:37 Transition towards autonomous AI agents for tasks.
00:01:18 TasCade emerges as a competitor with multi-AI agents.
00:01:53 TasCade offers a user-friendly drag-and-drop UI.
00:04:01 Potential for personalized AI filters for news curation.
00:05:20 Automation of content creation with AI agents.
00:07:19 Ra introduces powerful multimodal language models.
00:07:44 Ra Core competes with established AI models effectively.
00:08:33 Ra Core's capabilities in multimodal tasks and responses.
00:12:01 Application of AI in real-world tasks like robotics.
00:13:43 Ra Core's accurate analysis of complex video content.
00:16:10 Leveraging AI for business growth and innovation.
00:17:32 Rapid progress in AI models like GPT 4.
00:18:05 Testing and comparing AI models across various tasks.
00:23:55 Challenges in copyright laws for AI training data.
00:24:05 Proposed legislation on disclosing copyrighted data for AI.
00:25:02 Concerns about fair use and AI learning from data.
00:25:11 Political influences and interests in AI copyright laws.
00:25:25 Debate on AI models using copyrighted works.
00:25:55 Impact of large tech companies on AI regulations.
00:26:24 Significance of Google's response to copyright violations.
By YouSum
If I understood you correctly you are advocating using AI to reinforce your own bubble. This is a mighty double-edged sword
Good point, unfortunately a lot of algorithms already do this
I think what he mentions actually fights this! Imagine you getting a video you should know about based on your tilt! Might break your tilt! But also you might never recover!
isn't that just the internet and digitalization already? 90% of google is self-inflating bubble math.
Does anyone have eyes on catalyst? They are definitely working on it. Far runner dragonflies 1
bro did a face reveal to beat the ai allegations
Yeah 😂😂
He has shown his face on previous videos tho.
He's a reptiloid
Could still be ai
IOW, copyright is dead for everyone except the richest people in the world and Wes Roth is fine with that.
I can't imagine batching being any good for agents. Imagine having one thought every 24 hours, a simple task might take months to complete.
There's no question. Copyright isn't morally coherent. Copying is creation, not theft. Nobody has the (moral) right to PROHIBIT others from CREATING anything that isn't harmful.
It is, if you don't extend the COPY part of copyright to fair use. You shouldn't be allowed to just copy someone else work and make money on it, but you should be allowed to do so if you create something new using it.
@@cristianandrei5462 " You shouldn't be allowed to just copy someone else work and make money on it" Why? There is literally no reason behind this assertion. It's an unsubstantiated assumption, and it's wrong.
@@weksauce It's extremely counterproductive in my opinion to have people / businesses taking others intellectual work and just sell it while the original creator makes nothing. If you take into account that creating a song/ book/ a patente for an invention has costs, it's even worse, the one who copies gets the work that cost the other money and time for free and has the advantage in a free market, it just undercuts the original creator.
@@cristianandrei5462 There's no such thing as "intellectual work" nor "intellectual property". Copying something "intellectual" (an idea) INCREASES its value. Calculus is worth more the more people know it.
"If you take into account that creating a song/ book/ a patente for an invention has costs" So what? Just because something "has costs" doesn't mean it's property, or that one should be allowed to prohibit others from doing it! Punching myself in the face is costly; I don't see you arguing that if I do it (first) I ought to be able to prohibit others from doing it. This is equally stupid.
There's no meaningful distinction between "original" creator and "not-original" creator. They're all creators. Creation is fine. Creation in exchange for money is fine.
@@weksauce You sure seam to have a problem understanding simple logic, so I'll just give you an exemple and end it here. Let's say company A spends 100 milion dollars to do research and discover a new pharmaceutical drug, money it mostly gets from inverstors. Than company B just comes and produces the drug that company A discovered, without any research cost, and sells it on the market. Company A goes bankrupt, because it has to sell at a higher price, it has to pay 100 million to investors. In the end it proofs unprofitable for both inverstors and companies to do research...
Adam is for the people of the gov from the corps… if he can’t do anything no one can… no one can
I'd sign up for that ADHD agent! It can't be that hard to be honest. Stream screenshots and ask an LLM if the behaviour makes sense.
What ai builders need to be doing is building their ai agents around specific job roles. AI experts in job positions, that can communicate with one another.
That’s exactly what one would expect from the term ‘mixture of experts’. That’s where it’s heading
In other words design an AI replacement for humans?
They are trying, but the biggest problem of AI now is consistency. Gpt 4 is smart, the problem is its 10- 20 % of the time wrong, it misunderstanding the assignment... on complex tasks. Even worse, 1-5% of time it fails simple tasks. When you have a workflow with multiple (tens, hundreds) of "agents" (calls to a LLM) that compounds quickly. Even open source, small models are pretty capable, but in a real stakes situations (like dealing with costumers for a business, or doing important paperwork) they are not reliable enough (you need something that does the job close to 100% of the time, not randomly fail for whatever reason).
RBC is Royal Bank of Canada
So even if you not recording, it makes sense to me 🤔
Another great one, thanks Wes!
Simpie
Please place timestamps
If you replace human with machine(take away humans spending power). Who will buy the machine made products?
Nobody. Hence this is why people are talking about things like UBI.
@@pvanukoff For Americans or for the global population?
@@pvanukoffWith the talks of inflationary Government spending I wonder how that’s gonna even work. I guess this is something being accounted for (pun intended) in the new CBDC financial system that they plan on implementing.
Every artist is trained on others works
All these policymakers have no idae how large studios produce IP or music for that matter.
They have people that actively go out and look for uncopyrighted or easy to copy ideas and straight out recreate it with high production value.
On the low end, as i have stated multiple times on this channel, all creative courses actively teach you to 'get inspiration' by all means nessecary. What do you think a moodboard is for?
This constant misunderstanding of what the creative process really is like is mindblowing. People do not just manifest things into existence that is impossible. It would be evolutionary disadvantagous to produce ideas this way. You sample & copy ideas that work. You copy your parents and social group. That is how you learn. Producing ideas is a costly matter from a biological perspective.
The 3-Body Netflix series is pretty weak (Reader's Digest version), compared to the Cixin's mind-expanding concepts on the pages of his 'Remembrance of Earth's Past' trilogy. A must read for any human.
RBC = Royal Bank of Canada
at this point anyone left still making blog posts is doing it just for other AIs to read to make blog post for their humans
Watch 3 body problem from chinese people (San ti). Netflix one is just cheeky american culture flavored PC-filtered version with unrelated scenes.
That bill (at 25:00) is something the government may like, but the thinking behind it is epically short sighted. If they succeed in imposing restrictions on AI companies, they will simply move to a different country (aka Japan) and continue development there. Movement of a mega corpration (which comprises of 20-50 people) isn't may take a bit of work, and a bit of money, but a relocation could be a set back of a month or less in the modern world. Those imposing bills believe in authortarian control and actually think they are going to 'control' the development of AI technology. The reality is that they are simply going to lose control to another country (aka Japan). And the artists, who are screaming about unfair practices will still have any art they display on line still used for training... unless they collectively find a way to keep their art offline.
Complaining about the size of your parachute isn't terribly relevant after you jumped out of the plane. Copyright was a non enforcable issue ever since we could hit record on a casset tape and rip songs from the radio (1970-80s).
Lets not have these ai agent behind a $1000/month paywall when the wealth gap is already this large
in this AI hype and race that are happening for one year at least, I haven't felt any benefits yet. So everything is in early stage with no viable busines model.
It might be similar to crypto coins run that haven't gave any fruitful results for humanity just yet except ability for dictators to avoid sanctions.
Filtering to your interests would be nice, but what would be nicer is separating trash articles from good ones.
Cactus.
And it could be a team of 10 people who invent an AI that is causing a lot of trouble in future. This is why this thing is so dangerous.
Hide