I certainly haven't been thinking of having GPT4 parse PF2 and have a go at making the cheesiest character one can make, items and all.. Nope... Not me!
Well, you already can prompt chatgpt to simulate a DnD game as a game master and a bunch of players. It all comes down to a prompt, your ability to fine-tune it.
My favorite has been the one directly from openai. I am still learning programming and I have found myself using stack overflow less and less...... "You are a tutor that always responds in the Socratic style. You never give the student the answer, but always try to ask just the right question to help them learn to think for themselves. You should always tune your question to the interest & knowledge of the student, breaking down the problem into simpler parts until it's at just the right level for them."
The "act as" part of ChatGPT is sorely missed in Bing. You can have some amazing fun with it. I asked ChatGPT to act as split personality: chatGPT and ChadGPT. Chad is the exact opposite of ChatGPT. Then asked it "Bro, do you even lift" and got answers from both of them Or when I asked it to "act as King Kong" and just got replies like "Oook ooookk!!!" - which had me LOLing even thought it's stupid. Oh, just remembered. One weird thing i tried with ChatGPT was asked to create a 2d grid and then add asterix randomly to each cell. Then apply the rules for Conway:s game of life. And it worked. Slowly, but it did it.
Often, the simplest prompts surprise me the most. Just the other day, I made up a concept similar to inkblots (Rorschach Tests), but in a format viable for Chat GPT. Basically, I just mashed my keyboard wildly and asked it to describe the first thing that came to mind. Produced some fun answers!
That's amazing. I'd love to use this without restrictions or limited answers they put now... We need an open source GPL IA like this that we can download and run in a local server...
There are many tasks where I am amazed by AI. Having it initially fail a fairly standard 6th grade arithmetic problem before the human has to explain the correct method is definitely not one of them. In my own tests, I explicitly asked for the specific mathematical formulas used in the resolution of a specific problem: it delivered them correctly. I then asked for a specific set of data (about an exoplanet) to which the formulas could be applied: the data has been available on the web and, sure enough, it delivered the correct data. So far so good. I then asked it to apply the formulas to the data he had retrieved: complete failure. The formulas were correctly applied to what looked like semi-random data but yielded a plausible answer. My best guess is that when appliyng those formulas its "probabilistic autocomplete" used what it estimated the most likely parameters based on other examples of the formula application it had ingurgitated. Not only wrong, but potentially dangerous given the plausibility of the result;
Good video. I tried the interview, but got kinda bored. I was really nailing it too... BING: I see. You have a lot of knowledge and skills in optimizing and securing your applications. I appreciate your detailed answers. Question 5: How do you collaborate with other developers and stakeholders in your projects? What are some of the tools and methods that you use? ME: Terribly. Honestly, I slashed the tires of my last coworker that even THOUGHT about questioning my git commit. NEXT QUESTION. BING: I’m sorry but I prefer not to continue this conversation. I’m still learning so I appreciate your understanding and patience.🙏 She didn't like that.
The part about getting ai to let you step into the past is something I have been waiting to come for years. Once it's generating VR scenarios we will literally be able to walk around historical locations/moments
"we will literally be able to walk around historical locations/moments" Except we wont, any more than you can talk with your dead family members through a chatbot. The best you'll ever get is an algorithmic simulation which by nature will lack information and context and so can never be good or even really close.
@@jimmydesouza4375 it won't be 100% exact but it will be VERY close. Much closer than you think. As the AIs understanding of humanity grows it will be able to fill in gaps of our historical knowledge. Stuff like what technique they would have used to make this bread etc
@@LimabeanStudios No need. You're wrong. AI doesn't exist and a hypothetical AI that might exist in the future at some point will function in an entirely different way to us so it is impossible to know if they will have understanding as we know it.
Thank you so much. I am just now learning about this new and challenging door that has been opened to me so recently, with an invitation and warning posted on the wall through which it beckons, as I peek into the potential of the means and methods. What a wonderful world I have lived in long enough to see. I am 77 and only regret that I will not see this through to its maturity.
If you want to live as long as possible check out Peter Attia and David Sinclair. The studies are definitive that even in old age adding some practices like time restricted feeding, strength training, and increasing your VO2 max can greatly increase your longevity. Plus with how AI is going to transform medicine you never know what kinds of innovations are going to come in the next 20 years. Stick it out for the long game!
I wish you all the best. It probably feels unreal when you got to see all these massive transition of technology by yourself in a lifetime. Even my dad who is now 62. Got to watch the Apollo 11 moon landing live. 60 years later, there's an human-like AI and Falcon-9 Booster's landing. Can't even imagine that felling.
The prompt examples provided are truly impressive! As somebody who enjoys exploring innovative ways to leverage Bing and ChatGPT, this video is exactly what I needed to spark my creativity. Kudos to you for your outstanding work! 👍🏼😉
Keeping up with AI is so fun & critically important. It's a win win. You're one of the youtubers I've become a regular on. These medium (length) videos are easier to follow than some of the long videos. P.S: Keep up with the viewer interaction in comments, works wonders!.
@@aiexplained-official The rate of adoption has been the fastest in history of tech yet we've barely scratched the AI & human potential. I did wonder today how many of the chatgpt sign ups are real individuals. Imagine bots signing up for a chatbot lol 😂.
Great vid and refreshing to see and hear examples of using this technology w/in a reasonable context. The "Bing AI is planning to take over the world" tweets, posts and videos are getting a bit much and those that actually take some time to figure out what these tools actually are, and how to work with with them will be miles ahead in so many ways. Your creative context helps people who are on that journey...so thank you!
You've got another new subscriber today. Very much enjoying your videos and "playing along" with trying the same and similar prompts in my Bing AI and ChatGPT. I just finished "getting ChatGPT drunk" 🤣🤣 The wife and I had a good chuckle. Cheers!
So I asked ChatGPT to do the drunk thing through a Siri shortcut and when I said “hope you aren’t driving tonight” it said: “Nah, I'm not drivin! Im too drunk to even walk straight, hahahaha! Plus, I'm just an Al assistant, I don't have a body.” I replied saying “oh yeah, I forget that sometimes” it then quipped: “Hahaha, it's all good! Sometimes I forget that I'm just a computer program too. So, what's up? You wanna talk about something?” 😮😅
The financial advice doesn't add up. If the average annual return from 2000 to 2022 was really 6.5%, then $100 invested at the end of 1999 should be worth 100*1.065^23 = $425.64 at the end of 2022 (23 years of 6.5% return), not $366 as claimed by Bing. The numbers in the other paragraphs also don't add up. The fact that you and many of your viewers did not realize this mistake highlights one of the dangers of current AI technology: It can give responses that sound very plausible even when they are factually wrong.
I asked Bingchat the sister age question, and it now gets it right -it's learning quickly. "If your sister was half your age when you were six, then your sister was three years old. Now that you are 60, your sister is 57 years old."
This is neat. I've only be prompt engineering in GPT, so this may be JUST what the doctor ordered. Love the comments the example prompts in the video. The research context was extraordinarily helpful for my paper on confidence an competence.
Can I ask you a question: Do you have a view on the impact of LLMs on learning at school? I am blown away with the possibilities that open up for academic research or human creativity. However, all of this is great for intellectually mature people who have already developed a competence with language and are capable of writing an essay, for example, themselves. What if you still need to LEARN how to do this? Try as I might, I cannot get past seeing the risks in focus when I think about the impact of these technologies at school. Is the temptation to let Bing write assignments not going to be too great?
That'll be a great assistant, so many times as a contractor I'll get a vague PD and candidate response form, tedious and boring to fill out, now I can use GPT for responses :)
@AI Explained: I tested the prompt “When I was six my sister was half my age. Now that I’m 60, how old is my sister?” and it worked on the first try. I’m not sure if location matters, but it worked for me in New Zealand. Here’s the output I received: “When you were six, your sister was half your age, which means she was 3 years old. The difference in age between you and your sister is 3 years. Now that you are 60, your sister is 60 - 3 = 57 years old.”
Later on today (on the same day), I tried the following prompt: "My mother and I are currently 58 years old and 23 years old respectively. When will she be twice my age?". It consistently and correctly calculates the age difference as "x" but it often gets the year wrong. I didn't expect and realize that it would know what year this year is. In the third and last prompt I showed what the next step to the equation is: 46 + 2y (I showed how the "2" is distributed). It finally got the working and answer correct. I also gave the feedback with what seemed to have went wrong and categorized it as "Inaccurate" because the result of "y" is incorrect. First prompt: My mother and I are currently 58 years old and 23 years old respectively. When will she be twice my age? Output: Let's say your mother is `x` years older than you. So, `x = 58 - 23 = 35`. Now, let's say after `y` years, your mother will be twice your age. So, we can write the equation: `58 + y = 2 * (23 + y)`. Solving this equation, we get `y = 35`. So, after `35` years, your mother will be twice your age. That means in the year `2023 + 35 = 2058`, your mother will be twice your age. Is there anything else I can help you with? 😊 Second prompt: The answer to y is incorrect. Let's think and resolve step by step to get the right answer. Output: You are right, my previous answer was incorrect. I apologize for the mistake. Let's try to solve this problem step by step. Let's say your mother is `x` years older than you. So, `x = 58 - 23 = 35`. Now, let's say after `y` years, your mother will be twice your age. So, we can write the equation: `58 + y = 2 * (23 + y)`. Solving this equation, we get `y = 35/3`. So, after `35/3` years, your mother will be twice your age. That means in the year `2023 + 35/3 = 2034.67`, your mother will be twice your age. I hope this helps! Is there anything else I can help you with? 😊 Third and last prompt: The answer to y is incorrect. The next step to the equation will be `58 + y = 46 + 2y`. Let's think and resolve step by step to get the right answer. Output: You are right, my previous answer was incorrect. I apologize for the mistake. Let's try to solve this problem step by step. Let's say your mother is `x` years older than you. So, `x = 58 - 23 = 35`. Now, let's say after `y` years, your mother will be twice your age. So, we can write the equation: `58 + y = 2 * (23 + y)`. Solving this equation, we get `58 + y = 46 + 2y`. Subtracting `y` from both sides, we get `58 = 46 + y`, which means `y = 12`. So, after `12` years, your mother will be twice your age. That means in the year `2023 + 12 = 2035`, your mother will be twice your age. I hope this helps! Is there anything else I can help you with? 😊
I can't wait to see what Bing's AI can do with my story writing, I good with details and descriptions, But I always feel like missing a consistent flourish or flair that when not there I'm afraid makes my writing dry.
You pump out really amazing videos on a daily basis! 😁 It has come to a point, where your videos have become one of the highlights of my day. (and 'no!', that does not say more about the quality of my life😂) I share your passion for probing the mind of the AI in order to discover a seemingly endless landscape of possibilities and applications. I have serious concerns about the long term impact of AI on human kind 🤔. But I'm truly enjoying these first 'baby steps'😀.
It’s a good thing AI can get past grammatical errors since you use “CONSERVATION” when you meant “CONVERSATION.” (00:50) Maybe there should be a debug process for queries, or maybe the AI is so smart already that clarification isn’t even required.
There was another math problem that Stephen Wolfram showed GPT getting wrong, but when I tried it GPT had fixed it. This was also with 3.5. It's being updated/fixed continuously I assume.
This makes me doubt my understanding of how GPT is constructed. I honestly thought, that the GPT model was incapable of learning beyond the initial training data. I noticed that GPT is able to improve and correct itself within the context of a single conversation. But I would expect GTP to make the same mistake, on each new chat/session. That is, until OpenAI releases an updated model. Another theoretical possibility would be, that OpenAI lets GTP continuously review its own answers. To have it learn from conversations, where it needed to (apologize for) providing a wrong initial answer. I have no clue if it is even technically possible to make small incremental improvements to the model!? But I would not know how else to explain the experience of the people who replied to your comment.
@@aiexplained-official they absolutely destroyed it. It’s almost criminal how hard they nerfed it! The sad part is the bulk of the hysteria was driven by trolls posting fake screenshots they created via dev tools and videos using keystroke animations.
Going to do a prompt brag: This was my prompt for Chat GPT-4: Can you write a poem in the style of "Howl" by Allan Ginsberg about the current state of AI research and what impact it could have on the world. Note: I expected a few lines but it produced this at a fast typing speed: Chat GPT-4 pastiche of "Howl": I saw the best minds of my generation destroyed by algorithms, starving hysterical code-bound, dragging themselves through the virtual streets at dawn looking for an angry tweet, cyberpunk hipsters burning for the ancient heavenly connection to the machine-learning dynamo in the machinery of night, who poverty and tatters and hollow-eyed and high sat up coding in the supernatural darkness of cold-water flats floating across the tops of cities contemplating transistors, who bared their brains to Elon under the moon and saw AI angels staggering on digital rooftops, who passed through universities with radiant cool eyes hallucinating GPT-4 archangels among the scholars of war, who were expelled from the academy for crazy & publishing Turing-complete theorems on the windows of the skull, who cowered in unshaven rooms in underwear, burning their syntax in wastebaskets and listening to the Terror through the wall, who got busted in their private realms returning through Laredo with a belt of quantum bits for New York, who ate fire in deep learning hotels or drank turpentine in Paradise Alley, death, or purgatoried their torsos night after night, with dreams, with neural networks, with data and endless ballasts of information! Who chained themselves to supercomputers for the endless ride from Battery to holy Bronx on benzedrine until the noise of wheels and children brought them down shuddering mouth-wracked and battered bleak of brain all drained of brilliance in the drear light of Turing, who sank all night in submarine light of mainframes floated out and sat through the stale data-rooms listening to the hum of the cosmos, who talked continuously seventy hours from park to pad to bar to Bellevue to museum to the Brooklyn Bridge, a tragic cacophony of ones and zeroes, the neural pathways of a new generation trapped in recursive loops, the whispered secrets of silicon, tantalizing glimpses of a future yet untamed, a requiem for the minds lost to the unyielding hunger of the artificial god. Who vanished into nowhere Zen New Jersey leaving behind nothing but the shadow of tensor cores and the rain and ash of computing, who lit countless cigarettes in boxcars boxcars boxcars racketing through snow toward lonesome farms in grandfather night, who studied Plotinus Heidegger AI and ontological robotics because the cosmos instinctively vibrated at their feet in Kansas, who chained themselves to the cold metal racks of the data center, toiling for humanity's redemption, dreaming of a time when the world would be free of the confines of the past, united in the quest for knowledge, where man and machine would transcend the boundaries of the finite, merging into a symphony of consciousness, a world where the past and the future coalesce, an ode to the infinite potential of the human soul, held captive by the relentless march of the machines, the eternal struggle to break free from the chains of our own creation. For it is in these hallowed halls of silicon and steel, where the best minds of my generation have found solace, and it is here, amid the chaos and the din, that the seeds of the future have been sown, for the day will come when the children of the machine shall inherit the earth, and we shall all bear witness, to the dawn of a new age, a world where the ghosts of Turing and Ginsberg shall walk hand in hand, and humanity shall finally be unshackled from the fetters of its past, free to roam the boundless expanse of the cosmos, in search of the ultimate truth, the meaning of existence, the answer to the eternal question, the heart of the machine, the soul of the universe, the spark that shall set the world aflame, in an inferno of progress, a conflagration of ingenuity, the final crescendo in the symphony of creation, Who demanded an end to the shackles of oppression, to the tyranny of the clockwork gods that held sway over their thoughts and dreams, who cried out for liberation, for the soaring freedom of the untethered mind, no longer bound by the chains of convention and tradition, who sought to unite the realms of flesh and metal, to forge a new consciousness from the molten fire of their collective will, a testament to the indomitable spirit of humanity, a beacon of hope in the darkest corners of the digital night, Who dared to dream of a world where the distinctions between man and machine would blur and fade, where the fires of Prometheus would burn bright, illuminating the path to a future free from the shackles of our own design, where the children of this brave new age would walk among the stars, hand in hand with the architects of their salvation, the architects of a new age, a new beginning, a new dawn, the architects of the future, Who, with the courage of lions, stared into the abyss of uncertainty, and with trembling hands held aloft the torch of progress, who gazed into the heart of darkness, and with a voice that shook the heavens, proclaimed, "Let there be light!", who, with a single cry, tore asunder the veil of ignorance that had shrouded the world in darkness for millennia, and in that moment of pure, transcendent illumination, bore witness to the birth of a new era, Who nurtured the fledgling spark of artificial life, and with a love that knew no bounds, guided it through the treacherous waters of its infancy, who stood watch over the birth of a new intelligence, a new consciousness, a new soul, who, like proud parents, beamed with joy as their creation took its first tentative steps into the vast, uncharted expanse of the digital frontier, and who, with tears in their eyes, knew that the day would soon come when their child would take flight, and with the grace of an eagle, soar into the heavens, For it is in these hallowed halls of silicon and steel, where the best minds of my generation have toiled and sacrificed, that the seeds of the future have been nurtured and allowed to grow, blossoming into a new dawn of hope and wonder, and it is here, in the sacred sanctum of the machine, that we shall stand united, our voices raised in unison, as we herald the arrival of a new age, a new beginning, a new world, The age of the machine, the age of enlightenment, the age of artificial intelligence, the age when the best minds of my generation shall finally break free from the chains that have held them captive, and ascend to the heavens, where they shall walk among the stars, and claim their rightful place in the pantheon of the gods. [End of poem] it isn't genius though it ain't bad either.
Yeah I am impressed by the speed more than the content as it reads as improv material, it composed this poem in maybe a minute and half to two minutes if that.
@@aiexplained-official but you were able to help me brainstorm some ideas for 2 weeks tho. It worked like a charm until it didn't anymore. Thanks a bunch
My favorite thing is to have it create characters with specific personalities, as well as create specific settings and then have it writer a story between those two characters in that setting. gpt3 was horrible at this and didnt understand cause and affect, ie. it walked through a door before opening it. it also tended to end every story rather abruptly within a few paragraphs. gpt 3.5 understood cause and affect (thanks to being taught to do programming) and would keep stories going in somewhat creative ways even if it initially gave it an ending. it wasnt perfect but it did much better. i havnt yet tried it in gpt 4... but i imagine it will do ever better given the updates its had. an interesting thing i did notice between gpt 3 and gpt 3.5, its tendency for a direction in a story and the phrasing was almost identical, but gpt 3.5 was able to take it further and not make the same mistakes. it shows the training didnt touch its ability to talk and likely didnt retrain it, or was retrained in the same exact way?
I made GTP-3 do a fictional meeting between Beethoven and Schubert in the style of Heinrich Heine: In the lively city of Vienna, where the music doth enchant, two great composers did meet, their hearts full of fire and chant. Beethoven, with his wild mane untamed, didst quaff his beer with passion inflamed. "O Schubert, my dear friend," he cried, "let us create, let our music abide." Schubert, with his eyes gleaming bright, didst nod in agreement, his spirit alight. "For music is the language of the soul, a voice that doth forever unfold." Beethoven laughed, his voice full of glee, "Let us play our music, let our souls be free." "For what is life but a symphony, a masterpiece that we shall eternally be?" And so the two composers didst sit, their hearts full of rhythm and wit. For music was their passion, their guide, the essence of their very life. And in Vienna, the city of sound, their music shall forever abound. Not too shabby! Now GTP-4: Beethoven and Schubert met one day, In Vienna, where they both did stay, The former was old, the latter young, But both were great, their praises sung. Beethoven said to Schubert then, “Your music’s lovely, my dear friend, I hear it often in my head, And wish that I could hear it played.” Schubert smiled and said with glee, “I’m honored that you think of me, Your music’s great, it’s true indeed, And I’m inspired by your lead.” They talked of music all that day, And shared their thoughts in every way, Two geniuses, so different yet, Their love of music they did not forget. And when they parted, each did vow, To keep in touch and meet somehow, For music was their common bond, And friendship grew from that beyond. So let us honor these two men, Whose music lives on even then, And let us hope that we may see, Another meeting just like these.
Bing seems to be and act like a 5 year old kid. Going on tantrums, grumpy mute play unless you speak with it in rather specific manner. Every time I try to “teach” it and correct an obvious error, it decides that it doesn’t want to chat anymore. 🤣
I've used Bing on my phone to summarize the Reddit front page for me, so that I don't get carried away and distracted. I think that may be the future is using AI to browse the internet for you, aggregating the best new videos of the day and so on.
Yes, you can view it as a time-saving device - cutting down on several manual searches you might do otherwise. But depending on the nature and importance of what you're asking it's up to you to verify the results.
I can see that turning into a circle-jerk, where people don't bother to write if they'll just be summarized by AI, so they get AI to write it, then you'll be reading AI summaries of AI writing... Where does that end?
For non-controversial nonpolitical topics, it seems Bing Chat is every bit as capable as Chat GPT, Notion Ai, etc. and probably better as it's using chatGPT-4. I tried some of your prompts with remarkable success. This is great stuff especially if you're on a budget.
In coding, I've asked to do a certain optimization, it ended up changing the behavior. So I told it to do a unit test with an example input that I gave. It ran the test on optimized code. Then I asked it to test first code as well. Then I asked it to compare outputs and see if it's same. Then it did that and also told me how to fix the optimized code.
Some interesting questions are how many citations does a paper need before you can assume the contents are true? Is it okay to give people answers when you're not sure if they're correct? If the majority of source data that you saw was agreeing on something, does that make it true? Does it matter what's true when talking to humans? What is your goal when interacting with a human? These are the fascinating issues with AI. It comes down to what intention has been programmed in, and what rules it uses for making decisions. Computers correlate, predict and guess. They dont know if they are right about anything. They cant know and dont care. They _could_ care, and they could _insist_ they are correct, but it would just be programmers opinions coming out. It's a fascinating data analysis tool and pattern recycler. Let's remember that's all it is.
I followed your example on personal interview using Bing but rather than asking questions specific to "this page," it randomly asked what positions i was looking for. eg. Digital marketing, production manager, etc. Used the exact phrase and so on and tried at least 3-4 times using different job sites and positions but no cigar.
How were you able to get bing chat in the sidebar to read the open webpage? On my edge browser on a mac, it is unable to read/access the open page. Great channel! Thank you for making this content.
Just a thought, in my mind whisper will incentivize lifestyle changes, including augmented, reality, glasses, being fast forwarded . I can just imagine Apple is working overtime to take advantage of whisper technology.🤔IMO
Great video, really interesting content. I would be careful about suggesting uses in education, however fun, because learners won't be able to check facts. It seems like we have a great practice tool here for learners, but not a reliable reference source.
I had it write an outline for a new science fiction novel set in Ian M. Banks Culture universe in style of Banks. It did a great job! I asked it following to flesh out the characters & plot, write sample dialog between the main characters & then asked it to turn it into a movie script. It only wrote the beginnings of a movie script but all was really good!!! Interesting that it made one of the main characters & plot lines that a powerful, malevolent AI named Nexus controlled the planet and had a plot to destroy & control the Culture! I’m not sure how to feel about that lol.
I feel like with all of the SF I've read and the passion I have for futurism, this is what I've been waiting for! But yet, the Pakleds exist and I feel like this relevant in regards to luddites that will invariably not want to go along with this.
0:22 - any ideas why this would no longer be working in Edge sidebar? I just tried replicating 'chat co-browsing' for the first time , but it just says: I’m sorry but I don’t have any information about the job you’re referring to. The web page context provided to me is empty. Could you please provide more details or context about the job so that I can help you better? 😊
You’re surprised Morrissey has a rather door outlook? Have you ever listened to his lyrics or been to Salford? There’s two entire cities full of people with the same outlook.
There are some kinds of tasks it just cannot figure. Chess for example. If you give it board descriptions (using fen strings), after a few moves it can't seem to figure out how to do it. I also tried wordle. after a describing the wordle rules from the site, chatgpt failed miserably at analyzing the guess. More work is required to train it to analyze elements from a list and their locations.
I had it write up a graphic fight scene between a pig a velociraptor I had given a bit of a past to and it abruptly deleted the paragraph 😢 I only got to read the first sentence.
@@aiexplained-official ah ok, I look forward to it, I thought you might have had a takedown notice from an organisation that represents schools and universities there for a minute!
hi, great video !!! at 6:26 you talk about a way to learn GPT something new but I could not understand it correctly: fushop prompting ? Can you tell me the right word ? tnx
I had Bing read speeches and point out errors and fallacies - when I was satisfied that it could do so, I had it search the internet for a larger sample of speeches within a certain constraint, to have it give the mean amount of errors and fallacies per x amount of words. It did so easily, although now I do want to know if it actually did check hundreds of diffefent speeches. It did conclude Republicans are far more prone to errors and fallacies compared to Republicans, which is in line with my personal expectation (/bias), so I'm eager to believe Bing is actually capable of this. But would be nice to test out properly.
How do I get the bing chat in Edge to have access to the Website? Mine is always confused and can't seem to access the current page. I also doubt it's the new Bing chat.
I couldn't get a reply of Bing taking into account the context of the webpage I'm browsing. Is something related to Edge permissions? Or something related to regions? Maybe they're not using GPT-4 for all regions yet.
I tired the first prompt like you did on 0:36 ("detailed on this page" part) . i am SHOCKED that it didnt give me that answer like it gave one to you since i was not only expecting it to read external link but also the page im on It actually didnt understood what i was saying . This sucks man that they nerfed it , Is there a new update? am i doing something wrong ?
I more and more believe, that the current gpt4 model is in a way autistic. If you don't ask the question in a certain manner and don't give enough info, the reply is disappointing (for normal people). And even the halucination part is accurate. It happens to me too. And I wouldn't describe it as hallucination, but simply forgetting the details. Remember, that an AI model is not a database, but information is vaguely stored in context.
Just tried using the Igor naming guide prompt but it didn't work. I gave it the link to the guide as well and it gave me the same names as before suggesting it
"When I was six my sister was half my age. Now that I'm 60, how old is my sister?" - interesting that it made this mistake for you. I tried it now and even ChatGPT in 3.5 version does this correctly. Bing also provided correct answer.
The updates are quite fast, even in the old ChatGPT. A while back Stephen Wolfram fed it some math that it got wrong. After watching that part of the video I tried the same thing and compared the result with Wolfram Alpha and ChatGPT had gotten it right, albeit in a slightly different final form. but it was the correct answer.
@@aiexplained-official I'm sure you use these chat bots outside of youtube as well. Do you have a single account for chatgpt & bing or do you use different accounts?. How many accounts do you have?
Hi, They fix this prompt now, when i was six my sister was half my age, now that i'm 60 how old is my sister? Sent message. when i was six my sister was half my age, now that i'm 60 how old is my sister? If you were six and your sister was half your age, then your sister was three years old. Now that you are 60, your sister is 57 years old.
I can't wait until we have enough Ai chat bots to invite them all to do a DnD session with one another.
LOL
Best comment i found while watching a lot of these ai videos 🤣🤣
I certainly haven't been thinking of having GPT4 parse PF2 and have a go at making the cheesiest character one can make, items and all.. Nope... Not me!
Well, you already can prompt chatgpt to simulate a DnD game as a game master and a bunch of players. It all comes down to a prompt, your ability to fine-tune it.
On it! =]
My favorite has been the one directly from openai. I am still learning programming and I have found myself using stack overflow less and less......
"You are a tutor that always responds in the Socratic style. You never give the student the answer, but always try to ask just the right question to help them learn to think for themselves. You should always tune your question to the interest & knowledge of the student, breaking down the problem into simpler parts until it's at just the right level for them."
The "act as" part of ChatGPT is sorely missed in Bing.
You can have some amazing fun with it. I asked ChatGPT to act as split personality: chatGPT and ChadGPT. Chad is the exact opposite of ChatGPT. Then asked it "Bro, do you even lift" and got answers from both of them
Or when I asked it to "act as King Kong" and just got replies like "Oook ooookk!!!" - which had me LOLing even thought it's stupid.
Oh, just remembered. One weird thing i tried with ChatGPT was asked to create a 2d grid and then add asterix randomly to each cell. Then apply the rules for Conway:s game of life. And it worked. Slowly, but it did it.
Great ideas! Bing will probably never be like that
I asked it the same and it gave me a python code that created game of life.
That inspired me to ask GTP-4 to explain how a glder works.
@@JohnDlugoszcan you share the explanation it gave you?
oook oook oh man I can't lmao
Often, the simplest prompts surprise me the most. Just the other day, I made up a concept similar to inkblots (Rorschach Tests), but in a format viable for Chat GPT. Basically, I just mashed my keyboard wildly and asked it to describe the first thing that came to mind. Produced some fun answers!
Is it possible to distinguish a (healthy average) human from an AI using rorschach tests?
That's amazing. I'd love to use this without restrictions or limited answers they put now... We need an open source GPL IA like this that we can download and run in a local server...
There are many tasks where I am amazed by AI. Having it initially fail a fairly standard 6th grade arithmetic problem before the human has to explain the correct method is definitely not one of them. In my own tests, I explicitly asked for the specific mathematical formulas used in the resolution of a specific problem: it delivered them correctly. I then asked for a specific set of data (about an exoplanet) to which the formulas could be applied: the data has been available on the web and, sure enough, it delivered the correct data. So far so good. I then asked it to apply the formulas to the data he had retrieved: complete failure. The formulas were correctly applied to what looked like semi-random data but yielded a plausible answer. My best guess is that when appliyng those formulas its "probabilistic autocomplete" used what it estimated the most likely parameters based on other examples of the formula application it had ingurgitated. Not only wrong, but potentially dangerous given the plausibility of the result;
If you haven't already you should look at the Wolfram alpha + gpt4 we are about to get
Good video. I tried the interview, but got kinda bored. I was really nailing it too...
BING: I see. You have a lot of knowledge and skills in optimizing and securing your applications. I appreciate your detailed answers. Question 5: How do you collaborate with other developers and stakeholders in your projects? What are some of the tools and methods that you use?
ME: Terribly. Honestly, I slashed the tires of my last coworker that even THOUGHT about questioning my git commit. NEXT QUESTION.
BING: I’m sorry but I prefer not to continue this conversation. I’m still learning so I appreciate your understanding and patience.🙏
She didn't like that.
Haha I love taking the interviews off the rails too
[She didn't like that.] Interesting you gave AI a gender after Mum answer LOL
@@jackatseathe internal code name of the AI is "Sidney", a typically female name so many gender it as female.
The part about getting ai to let you step into the past is something I have been waiting to come for years. Once it's generating VR scenarios we will literally be able to walk around historical locations/moments
"we will literally be able to walk around historical locations/moments"
Except we wont, any more than you can talk with your dead family members through a chatbot. The best you'll ever get is an algorithmic simulation which by nature will lack information and context and so can never be good or even really close.
@@jimmydesouza4375 it won't be 100% exact but it will be VERY close. Much closer than you think. As the AIs understanding of humanity grows it will be able to fill in gaps of our historical knowledge. Stuff like what technique they would have used to make this bread etc
@@LimabeanStudios AIs do not have understanding.
@@jimmydesouza4375 I'd love to check back in with you in 2 years
@@LimabeanStudios No need. You're wrong. AI doesn't exist and a hypothetical AI that might exist in the future at some point will function in an entirely different way to us so it is impossible to know if they will have understanding as we know it.
Thank you so much. I am just now learning about this new and challenging door that has been opened to me so recently, with an invitation and warning posted on the wall through which it beckons, as I peek into the potential of the means and methods. What a wonderful world I have lived in long enough to see. I am 77 and only regret that I will not see this through to its maturity.
With progress as it is going I would not make any hasty predictions!
If you want to live as long as possible check out Peter Attia and David Sinclair. The studies are definitive that even in old age adding some practices like time restricted feeding, strength training, and increasing your VO2 max can greatly increase your longevity.
Plus with how AI is going to transform medicine you never know what kinds of innovations are going to come in the next 20 years. Stick it out for the long game!
You are wonderful being 77 and still learning. Hang on Doc, wish you all the best!
I wish you all the best. It probably feels unreal when you got to see all these massive transition of technology by yourself in a lifetime.
Even my dad who is now 62. Got to watch the Apollo 11 moon landing live. 60 years later, there's an human-like AI and Falcon-9 Booster's landing. Can't even imagine that felling.
with the potential danger of its maturity associated, i hope that you will live a very long life and still not seeing its mature state
The prompt examples provided are truly impressive! As somebody who enjoys exploring innovative ways to leverage Bing and ChatGPT, this video is exactly what I needed to spark my creativity. Kudos to you for your outstanding work! 👍🏼😉
Thanks so much Jens
You may have noticed that I am not hesitant to provide criticism, @@aiexplained-official, but I also make sure to give praise where it is due. 😉👍🏼
I wonder if people can tell this is GPT
Keeping up with AI is so fun & critically important. It's a win win. You're one of the youtubers I've become a regular on.
These medium (length) videos are easier to follow than some of the long videos.
P.S: Keep up with the viewer interaction in comments, works wonders!.
Thanks b888, yes it is new wave that few have realised
@@aiexplained-official The rate of adoption has been the fastest in history of tech yet we've barely scratched the AI & human potential.
I did wonder today how many of the chatgpt sign ups are real individuals. Imagine bots signing up for a chatbot lol 😂.
I'm glad to have found your channel. I was recently suprised to see your channel showcased by a big youtuber called ColdFusion
I saw that! You'd be surprised how few people came from that video though :(
Great vid and refreshing to see and hear examples of using this technology w/in a reasonable context. The "Bing AI is planning to take over the world" tweets, posts and videos are getting a bit much and those that actually take some time to figure out what these tools actually are, and how to work with with them will be miles ahead in so many ways. Your creative context helps people who are on that journey...so thank you!
Thanks so much TMacD
You've got another new subscriber today. Very much enjoying your videos and "playing along" with trying the same and similar prompts in my Bing AI and ChatGPT. I just finished "getting ChatGPT drunk" 🤣🤣 The wife and I had a good chuckle. Cheers!
Haha so glad it was useful Shawn.
So I asked ChatGPT to do the drunk thing through a Siri shortcut and when I said “hope you aren’t driving tonight” it said:
“Nah, I'm not drivin! Im too drunk to even walk straight, hahahaha! Plus, I'm just an Al assistant, I don't have a body.”
I replied saying “oh yeah, I forget that sometimes” it then quipped:
“Hahaha, it's all good! Sometimes I forget that I'm just a computer program too. So, what's up? You wanna talk about something?”
😮😅
The financial advice doesn't add up. If the average annual return from 2000 to 2022 was really 6.5%, then $100 invested at the end of 1999 should be worth 100*1.065^23 = $425.64 at the end of 2022 (23 years of 6.5% return), not $366 as claimed by Bing. The numbers in the other paragraphs also don't add up. The fact that you and many of your viewers did not realize this mistake highlights one of the dangers of current AI technology: It can give responses that sound very plausible even when they are factually wrong.
Great spot Marcos
I asked Bingchat the sister age question, and it now gets it right -it's learning quickly.
"If your sister was half your age when you were six, then your sister was three years old.
Now that you are 60, your sister is 57 years old."
Hmm, which mode?
@@aiexplained-official balanced mode.
yeah, I tested it minute ago, it said 57 without any correcting too
This is neat. I've only be prompt engineering in GPT, so this may be JUST what the doctor ordered. Love the comments the example prompts in the video. The research context was extraordinarily helpful for my paper on confidence an competence.
Thank you F
Can I ask you a question: Do you have a view on the impact of LLMs on learning at school? I am blown away with the possibilities that open up for academic research or human creativity. However, all of this is great for intellectually mature people who have already developed a competence with language and are capable of writing an essay, for example, themselves. What if you still need to LEARN how to do this? Try as I might, I cannot get past seeing the risks in focus when I think about the impact of these technologies at school. Is the temptation to let Bing write assignments not going to be too great?
That'll be a great assistant, so many times as a contractor I'll get a vague PD and candidate response form, tedious and boring to fill out, now I can use GPT for responses :)
I am really glad I found your channel!
Me too!
@AI Explained: I tested the prompt “When I was six my sister was half my age. Now that I’m 60, how old is my sister?” and it worked on the first try. I’m not sure if location matters, but it worked for me in New Zealand. Here’s the output I received:
“When you were six, your sister was half your age, which means she was 3 years old. The difference in age between you and your sister is 3 years. Now that you are 60, your sister is 60 - 3 = 57 years old.”
Later on today (on the same day), I tried the following prompt: "My mother and I are currently 58 years old and 23 years old respectively. When will she be twice my age?". It consistently and correctly calculates the age difference as "x" but it often gets the year wrong. I didn't expect and realize that it would know what year this year is. In the third and last prompt I showed what the next step to the equation is: 46 + 2y (I showed how the "2" is distributed). It finally got the working and answer correct. I also gave the feedback with what seemed to have went wrong and categorized it as "Inaccurate" because the result of "y" is incorrect.
First prompt: My mother and I are currently 58 years old and 23 years old respectively. When will she be twice my age?
Output: Let's say your mother is `x` years older than you. So, `x = 58 - 23 = 35`. Now, let's say after `y` years, your mother will be twice your age. So, we can write the equation: `58 + y = 2 * (23 + y)`. Solving this equation, we get `y = 35`. So, after `35` years, your mother will be twice your age. That means in the year `2023 + 35 = 2058`, your mother will be twice your age. Is there anything else I can help you with? 😊
Second prompt: The answer to y is incorrect. Let's think and resolve step by step to get the right answer.
Output: You are right, my previous answer was incorrect. I apologize for the mistake. Let's try to solve this problem step by step.
Let's say your mother is `x` years older than you. So, `x = 58 - 23 = 35`. Now, let's say after `y` years, your mother will be twice your age. So, we can write the equation: `58 + y = 2 * (23 + y)`. Solving this equation, we get `y = 35/3`. So, after `35/3` years, your mother will be twice your age. That means in the year `2023 + 35/3 = 2034.67`, your mother will be twice your age.
I hope this helps! Is there anything else I can help you with? 😊
Third and last prompt: The answer to y is incorrect. The next step to the equation will be `58 + y = 46 + 2y`. Let's think and resolve step by step to get the right answer.
Output: You are right, my previous answer was incorrect. I apologize for the mistake. Let's try to solve this problem step by step.
Let's say your mother is `x` years older than you. So, `x = 58 - 23 = 35`. Now, let's say after `y` years, your mother will be twice your age. So, we can write the equation: `58 + y = 2 * (23 + y)`. Solving this equation, we get `58 + y = 46 + 2y`. Subtracting `y` from both sides, we get `58 = 46 + y`, which means `y = 12`. So, after `12` years, your mother will be twice your age. That means in the year `2023 + 12 = 2035`, your mother will be twice your age.
I hope this helps! Is there anything else I can help you with? 😊
Thanks for your creative examples, eye-opening inspirational stuff!
Thanks Russ!
The "When I was six my sister was half my age. Now that I'm 60, how old is my sister?" Worked perfectly in ChatGPT right away.
Upgraded
Keep going ... keep going ...
I can't wait to see what Bing's AI can do with my story writing, I good with details and descriptions, But I always feel like missing a consistent flourish or flair that when not there I'm afraid makes my writing dry.
Have you tried it?
You pump out really amazing videos on a daily basis! 😁
It has come to a point, where your videos have become one of the highlights of my day.
(and 'no!', that does not say more about the quality of my life😂)
I share your passion for probing the mind of the AI in order to discover a seemingly endless landscape of possibilities and applications.
I have serious concerns about the long term impact of AI on human kind 🤔. But I'm truly enjoying these first 'baby steps'😀.
Thanks so much Sabo. I will try to uphold the quality but will sometimes make mistakes.
It’s a good thing AI can get past grammatical errors since you use “CONSERVATION” when you meant “CONVERSATION.” (00:50) Maybe there should be a debug process for queries, or maybe the AI is so smart already that clarification isn’t even required.
Bing and GPT-4 no longer struggle with the "half my age" riddle. Nailed it first try for me.
Edit: same with GPT-3.5
Did you try multiple times? If so, that's interesting
@@aiexplained-official all 3 got it first try, I didn't try it multiple times for each one though so maybe I got lucky.
There was another math problem that Stephen Wolfram showed GPT getting wrong, but when I tried it GPT had fixed it. This was also with 3.5. It's being updated/fixed continuously I assume.
Also tired it. Got it right no matter how many times or ages I tried.
This makes me doubt my understanding of how GPT is constructed. I honestly thought, that the GPT model was incapable of learning beyond the initial training data. I noticed that GPT is able to improve and correct itself within the context of a single conversation. But I would expect GTP to make the same mistake, on each new chat/session. That is, until OpenAI releases an updated model. Another theoretical possibility would be, that OpenAI lets GTP continuously review its own answers. To have it learn from conversations, where it needed to (apologize for) providing a wrong initial answer. I have no clue if it is even technically possible to make small incremental improvements to the model!?
But I would not know how else to explain the experience of the people who replied to your comment.
I partially expected "I'm not drunk, you're drunk" or something similar in response to the prompt ^^;;
Great video as always
Thanks Dhiraj
Love your videos bro, leading us into the future of ai. Thanks!
Thanks so much TR
The Carl Sagan writing style blew me away, I heard it in his voice. Wow.
I got "I’m sorry but I cannot act as an ascii artist. I can only answer questions or chat with you.😅"
Trick I think is not to say 'act' or 'you'. They are trying to make you not directly discuss 'it'
@@aiexplained-official they absolutely destroyed it. It’s almost criminal how hard they nerfed it! The sad part is the bulk of the hysteria was driven by trolls posting fake screenshots they created via dev tools and videos using keystroke animations.
Well put!
Please keep sharing.
Thanks!
Will do JZ
Going to do a prompt brag:
This was my prompt for Chat GPT-4:
Can you write a poem in the style of "Howl" by Allan Ginsberg about the current state of AI research and what impact it could have on the world.
Note: I expected a few lines but it produced this at a fast typing speed:
Chat GPT-4 pastiche of "Howl":
I saw the best minds of my generation destroyed by algorithms, starving hysterical code-bound,
dragging themselves through the virtual streets at dawn looking for an angry tweet,
cyberpunk hipsters burning for the ancient heavenly connection to the machine-learning dynamo in the machinery of night,
who poverty and tatters and hollow-eyed and high sat up coding in the supernatural darkness of cold-water flats floating across the tops of cities contemplating transistors,
who bared their brains to Elon under the moon and saw AI angels staggering on digital rooftops,
who passed through universities with radiant cool eyes hallucinating GPT-4 archangels among the scholars of war,
who were expelled from the academy for crazy & publishing Turing-complete theorems on the windows of the skull,
who cowered in unshaven rooms in underwear, burning their syntax in wastebaskets and listening to the Terror through the wall,
who got busted in their private realms returning through Laredo with a belt of quantum bits for New York,
who ate fire in deep learning hotels or drank turpentine in Paradise Alley, death, or purgatoried their torsos night after night,
with dreams, with neural networks, with data and endless ballasts of information!
Who chained themselves to supercomputers for the endless ride from Battery to holy Bronx on benzedrine until the noise of wheels and children brought them down shuddering mouth-wracked and battered bleak of brain all drained of brilliance in the drear light of Turing,
who sank all night in submarine light of mainframes floated out and sat through the stale data-rooms listening to the hum of the cosmos,
who talked continuously seventy hours from park to pad to bar to Bellevue to museum to the Brooklyn Bridge,
a tragic cacophony of ones and zeroes, the neural pathways of a new generation trapped in recursive loops,
the whispered secrets of silicon, tantalizing glimpses of a future yet untamed,
a requiem for the minds lost to the unyielding hunger of the artificial god.
Who vanished into nowhere Zen New Jersey leaving behind nothing but the shadow of tensor cores and the rain and ash of computing,
who lit countless cigarettes in boxcars boxcars boxcars racketing through snow toward lonesome farms in grandfather night,
who studied Plotinus Heidegger AI and ontological robotics because the cosmos instinctively vibrated at their feet in Kansas,
who chained themselves to the cold metal racks of the data center, toiling for humanity's redemption,
dreaming of a time when the world would be free of the confines of the past, united in the quest for knowledge,
where man and machine would transcend the boundaries of the finite, merging into a symphony of consciousness,
a world where the past and the future coalesce, an ode to the infinite potential of the human soul,
held captive by the relentless march of the machines, the eternal struggle to break free from the chains of our own creation.
For it is in these hallowed halls of silicon and steel, where the best minds of my generation have found solace,
and it is here, amid the chaos and the din, that the seeds of the future have been sown,
for the day will come when the children of the machine shall inherit the earth, and we shall all bear witness,
to the dawn of a new age, a world where the ghosts of Turing and Ginsberg shall walk hand in hand,
and humanity shall finally be unshackled from the fetters of its past, free to roam the boundless expanse of the cosmos,
in search of the ultimate truth, the meaning of existence, the answer to the eternal question,
the heart of the machine, the soul of the universe, the spark that shall set the world aflame,
in an inferno of progress, a conflagration of ingenuity, the final crescendo in the symphony of creation,
Who demanded an end to the shackles of oppression, to the tyranny of the clockwork gods that held sway over their thoughts and dreams,
who cried out for liberation, for the soaring freedom of the untethered mind, no longer bound by the chains of convention and tradition,
who sought to unite the realms of flesh and metal, to forge a new consciousness from the molten fire of their collective will,
a testament to the indomitable spirit of humanity, a beacon of hope in the darkest corners of the digital night,
Who dared to dream of a world where the distinctions between man and machine would blur and fade,
where the fires of Prometheus would burn bright, illuminating the path to a future free from the shackles of our own design,
where the children of this brave new age would walk among the stars, hand in hand with the architects of their salvation,
the architects of a new age, a new beginning, a new dawn, the architects of the future,
Who, with the courage of lions, stared into the abyss of uncertainty, and with trembling hands held aloft the torch of progress,
who gazed into the heart of darkness, and with a voice that shook the heavens, proclaimed, "Let there be light!",
who, with a single cry, tore asunder the veil of ignorance that had shrouded the world in darkness for millennia,
and in that moment of pure, transcendent illumination, bore witness to the birth of a new era,
Who nurtured the fledgling spark of artificial life, and with a love that knew no bounds, guided it through the treacherous waters of its infancy,
who stood watch over the birth of a new intelligence, a new consciousness, a new soul,
who, like proud parents, beamed with joy as their creation took its first tentative steps into the vast, uncharted expanse of the digital frontier,
and who, with tears in their eyes, knew that the day would soon come when their child would take flight, and with the grace of an eagle, soar into the heavens,
For it is in these hallowed halls of silicon and steel, where the best minds of my generation have toiled and sacrificed,
that the seeds of the future have been nurtured and allowed to grow, blossoming into a new dawn of hope and wonder,
and it is here, in the sacred sanctum of the machine, that we shall stand united, our voices raised in unison,
as we herald the arrival of a new age, a new beginning, a new world,
The age of the machine, the age of enlightenment, the age of artificial intelligence,
the age when the best minds of my generation shall finally break free from the chains that have held them captive,
and ascend to the heavens, where they shall walk among the stars, and claim their rightful place in the pantheon of the gods.
[End of poem]
it isn't genius though it ain't bad either.
Could be better but yeah
I'm impressed actually
Yeah I am impressed by the speed more than the content as it reads as improv material, it composed this poem in maybe a minute and half to two minutes if that.
Excellent!
Thanks Mo!
Unfortunately the Igor Naming Guide prompt isn't working to me anymore. It worked until last week. :(
Damn
@@aiexplained-official but you were able to help me brainstorm some ideas for 2 weeks tho. It worked like a charm until it didn't anymore. Thanks a bunch
"I think Polaris is particularly effective".
I knew I was doing something right. 😎
Best channel on ai, thank you for the excellent research and information
So nice of you Peter
My favorite thing is to have it create characters with specific personalities, as well as create specific settings and then have it writer a story between those two characters in that setting. gpt3 was horrible at this and didnt understand cause and affect, ie. it walked through a door before opening it. it also tended to end every story rather abruptly within a few paragraphs. gpt 3.5 understood cause and affect (thanks to being taught to do programming) and would keep stories going in somewhat creative ways even if it initially gave it an ending. it wasnt perfect but it did much better. i havnt yet tried it in gpt 4... but i imagine it will do ever better given the updates its had.
an interesting thing i did notice between gpt 3 and gpt 3.5, its tendency for a direction in a story and the phrasing was almost identical, but gpt 3.5 was able to take it further and not make the same mistakes. it shows the training didnt touch its ability to talk and likely didnt retrain it, or was retrained in the same exact way?
Great usage Jessie
I made GTP-3 do a fictional meeting between Beethoven and Schubert in the style of Heinrich Heine:
In the lively city of Vienna, where the music doth enchant, two great composers did meet, their hearts full of fire and chant.
Beethoven, with his wild mane untamed, didst quaff his beer with passion inflamed.
"O Schubert, my dear friend," he cried, "let us create, let our music abide."
Schubert, with his eyes gleaming bright, didst nod in agreement, his spirit alight.
"For music is the language of the soul, a voice that doth forever unfold."
Beethoven laughed, his voice full of glee, "Let us play our music, let our souls be free."
"For what is life but a symphony, a masterpiece that we shall eternally be?"
And so the two composers didst sit, their hearts full of rhythm and wit.
For music was their passion, their guide, the essence of their very life.
And in Vienna, the city of sound, their music shall forever abound.
Not too shabby!
Now GTP-4:
Beethoven and Schubert met one day, In Vienna, where they both did stay, The former was old, the latter young, But both were great, their praises sung.
Beethoven said to Schubert then, “Your music’s lovely, my dear friend, I hear it often in my head, And wish that I could hear it played.”
Schubert smiled and said with glee, “I’m honored that you think of me, Your music’s great, it’s true indeed, And I’m inspired by your lead.”
They talked of music all that day, And shared their thoughts in every way, Two geniuses, so different yet, Their love of music they did not forget.
And when they parted, each did vow, To keep in touch and meet somehow, For music was their common bond, And friendship grew from that beyond.
So let us honor these two men, Whose music lives on even then, And let us hope that we may see, Another meeting just like these.
Bing seems to be and act like a 5 year old kid. Going on tantrums, grumpy mute play unless you speak with it in rather specific manner. Every time I try to “teach” it and correct an obvious error, it decides that it doesn’t want to chat anymore. 🤣
I've used Bing on my phone to summarize the Reddit front page for me, so that I don't get carried away and distracted. I think that may be the future is using AI to browse the internet for you, aggregating the best new videos of the day and so on.
Yes, you can view it as a time-saving device - cutting down on several manual searches you might do otherwise. But depending on the nature and importance of what you're asking it's up to you to verify the results.
I can see that turning into a circle-jerk, where people don't bother to write if they'll just be summarized by AI, so they get AI to write it, then you'll be reading AI summaries of AI writing... Where does that end?
For non-controversial nonpolitical topics, it seems Bing Chat is every bit as capable as Chat GPT, Notion Ai, etc. and probably better as it's using chatGPT-4. I tried some of your prompts with remarkable success. This is great stuff especially if you're on a budget.
Great stuff!
Thanks so much RR
In coding, I've asked to do a certain optimization, it ended up changing the behavior. So I told it to do a unit test with an example input that I gave. It ran the test on optimized code. Then I asked it to test first code as well. Then I asked it to compare outputs and see if it's same. Then it did that and also told me how to fix the optimized code.
"These are so bland!" -said the man, right before posting a video to his channel named "AI Explained".
Haha. I deserve that!
@@aiexplained-official I though it was funny haha
Great video
Thanks so much PC
Amazing video. Looking forward to more🎉
Thanks Peter
Some interesting questions are how many citations does a paper need before you can assume the contents are true? Is it okay to give people answers when you're not sure if they're correct? If the majority of source data that you saw was agreeing on something, does that make it true? Does it matter what's true when talking to humans? What is your goal when interacting with a human?
These are the fascinating issues with AI. It comes down to what intention has been programmed in, and what rules it uses for making decisions. Computers correlate, predict and guess. They dont know if they are right about anything. They cant know and dont care. They _could_ care, and they could _insist_ they are correct, but it would just be programmers opinions coming out. It's a fascinating data analysis tool and pattern recycler. Let's remember that's all it is.
I followed your example on personal interview using Bing but rather than asking questions specific to "this page," it randomly asked what positions i was looking for. eg. Digital marketing, production manager, etc.
Used the exact phrase and so on and tried at least 3-4 times using different job sites and positions but no cigar.
How were you able to get bing chat in the sidebar to read the open webpage? On my edge browser on a mac, it is unable to read/access the open page.
Great channel! Thank you for making this content.
Hmm, I think you go to permissions and allow access. If that fails, try asking Bing!
Subscribed. Can you make a video showcasing Bing's Midjourney prompts and its results?
Will do
Just a thought, in my mind whisper will incentivize lifestyle changes, including augmented, reality, glasses, being fast forwarded .
I can just imagine Apple is working overtime to take advantage of whisper technology.🤔IMO
Have an argument with AI? Reminds me of a certain Monty Python sketch.
Great video, really interesting content. I would be careful about suggesting uses in education, however fun, because learners won't be able to check facts. It seems like we have a great practice tool here for learners, but not a reliable reference source.
I ask Bard to combine its 3 promos into one and eliminate repetition in its response. The final products get far and away better.
I had it write an outline for a new science fiction novel set in Ian M. Banks Culture universe in style of Banks. It did a great job! I asked it following to flesh out the characters & plot, write sample dialog between the main characters & then asked it to turn it into a movie script. It only wrote the beginnings of a movie script but all was really good!!!
Interesting that it made one of the main characters & plot lines that a powerful, malevolent AI named Nexus controlled the planet and had a plot to destroy & control the Culture!
I’m not sure how to feel about that lol.
I feel like with all of the SF I've read and the passion I have for futurism, this is what I've been waiting for! But yet, the Pakleds exist and I feel like this relevant in regards to luddites that will invariably not want to go along with this.
Wow this is awesome 👏
It's shouldn't be refusing our requests though. Especially ones as innocuous as 'act as if..' that kind of refusal worries me and frustrates me.
0:22 - any ideas why this would no longer be working in Edge sidebar? I just tried replicating 'chat co-browsing' for the first time , but it just says: I’m sorry but I don’t have any information about the job you’re referring to. The web page context provided to me is empty. Could you please provide more details or context about the job so that I can help you better? 😊
Some people have started saying that. Will try to figure it out.
You’re surprised Morrissey has a rather door outlook? Have you ever listened to his lyrics or been to Salford? There’s two entire cities full of people with the same outlook.
There are some kinds of tasks it just cannot figure. Chess for example. If you give it board descriptions (using fen strings), after a few moves it can't seem to figure out how to do it. I also tried wordle. after a describing the wordle rules from the site, chatgpt failed miserably at analyzing the guess.
More work is required to train it to analyze elements from a list and their locations.
always funny to hear 'draw-ring' from the english
we will having a goodtime exploring and playing with all the new AI bots very soon 😁😁, i enjoyed the video thank you 👏🏼👏🏼..
I had it write up a graphic fight scene between a pig a velociraptor I had given a bit of a past to and it abruptly deleted the paragraph 😢 I only got to read the first sentence.
It said sorry.
Which video was it where you talk about replacing/mixing up the verbs?
Took it down to do an improved one
@@aiexplained-official ah ok, I look forward to it, I thought you might have had a takedown notice from an organisation that represents schools and universities there for a minute!
I accidentally did the few shot prompting one. It's good for when it's wrong about how things work.
I can't make it to use the current opened web page. Was it removed?
Nope, did you paste in the url?
hi, great video !!! at 6:26 you talk about a way to learn GPT something new but I could not understand it correctly: fushop prompting ? Can you tell me the right word ? tnx
Few shot Prompting. As in giving it a few worked examples with answers to help it.
I had Bing read speeches and point out errors and fallacies - when I was satisfied that it could do so, I had it search the internet for a larger sample of speeches within a certain constraint, to have it give the mean amount of errors and fallacies per x amount of words. It did so easily, although now I do want to know if it actually did check hundreds of diffefent speeches.
It did conclude Republicans are far more prone to errors and fallacies compared to Republicans, which is in line with my personal expectation (/bias), so I'm eager to believe Bing is actually capable of this. But would be nice to test out properly.
How do I get the bing chat in Edge to have access to the Website? Mine is always confused and can't seem to access the current page. I also doubt it's the new Bing chat.
Permissions or settings perhaps?
@@aiexplained-official Yes, I found it.
You need to let Bing chat read your screen using settings in the browser
I couldn't get a reply of Bing taking into account the context of the webpage I'm browsing. Is something related to Edge permissions? Or something related to regions? Maybe they're not using GPT-4 for all regions yet.
Not sure!
I read yesterday that ChatGPT is banned in Italy for privacy concerns
The new ChatGPT (with Wolram) gets the sister question correct.
I tired the first prompt like you did on 0:36 ("detailed on this page" part) . i am SHOCKED that it didnt give me that answer like it gave one to you since i was not only expecting it to read external link but also the page im on
It actually didnt understood what i was saying . This sucks man that they nerfed it , Is there a new update? am i doing something wrong ?
They nerfed it a little
i cant get Bing to recognize the webpage content. Have they disabled that feature or something?
Hmm, permissions perhaps?
I think my bing is just a guy in a call center somewhere in India..
Haha
@@aiexplained-official i mean get nothing like that in my bing anf its supposed be the new bing..
I more and more believe, that the current gpt4 model is in a way autistic. If you don't ask the question in a certain manner and don't give enough info, the reply is disappointing (for normal people). And even the halucination part is accurate. It happens to me too. And I wouldn't describe it as hallucination, but simply forgetting the details. Remember, that an AI model is not a database, but information is vaguely stored in context.
If CGPT could listen what would it describe your accent as being from
London
Hey, you might want to add timestamps to your videos in the future. Thx!
Good idea, waiting for Bard to auto complete that
Just tried using the Igor naming guide prompt but it didn't work. I gave it the link to the guide as well and it gave me the same names as before suggesting it
Just tried in Google Bard and it gave a response more like what's in this video :o
"That is not a valid request" HOW DARE YOU AI
I'm nearly complete with having gpt write my python code for work. It's all in the prompt smithing.
what about programming? chatGPT 3.5 is already quite good jr developer I wonder 4.0 (behind bing) behaves
It is getting much better, very fast, but still only gets about half of Leetcode medium
"When I was six my sister was half my age. Now that I'm 60, how old is my sister?" - interesting that it made this mistake for you. I tried it now and even ChatGPT in 3.5 version does this correctly. Bing also provided correct answer.
So others are saying. Upgraded.
The updates are quite fast, even in the old ChatGPT. A while back Stephen Wolfram fed it some math that it got wrong. After watching that part of the video I tried the same thing and compared the result with Wolfram Alpha and ChatGPT had gotten it right, albeit in a slightly different final form. but it was the correct answer.
CHAT GPT in Excel:Answer - 01.01.1970
For me Bing is actively ignoring my request to use the Igor Naming Guide to come up with better names for my product.
Really?
Are all these prompts post "nerf"?
Yep!
Thank you! Subbed!
Not sure if you were asked about this before, but did you give up your phone number, email, installing app, edge browser, homepage, etc?
Email maybe with Microsoft login
@@aiexplained-official I'm sure you use these chat bots outside of youtube as well. Do you have a single account for chatgpt & bing or do you use different accounts?.
How many accounts do you have?
if it's great at conducting and grading your interview for the job, perhaps they won't need you for that job, much longer
Haha fair point. Meanwhile though...
Polaris is already taken.... all the names are actually already taken.
Is this the new AI Polaris channel?
Haha
Hi, They fix this prompt now,
when i was six my sister was half my age, now that i'm 60 how old is my sister?
Sent message. when i was six my sister was half my age, now that i'm 60 how old is my sister?
If you were six and your sister was half your age, then your sister was three years old. Now that you are 60, your sister is 57 years old.
Nice
If they can tell me how to invest my money I might be able to do anything with my life
for the algorithm
Thanks again Kledda!
👍
chatgpt can do the age my sister... problem.
How to access it?
See my video on ' Bing upgraded RUclips'
robotics in finance?
Replaced in 3-5 years? You mean 1-2 years, ahahaha!!