GPT: The Second Renaissance

Everyone knows that GPT and AI are going to change the world, but HOW? I have a few ideas to share with you that I've not seen elsewhere and it start with Beer.
❤️ If you would like to support what I do, I have set up a patreon here: / noboilerplate - Thank you!
📄 All my videos are built in markdown, transcript sourcecode available here github.com/0atman/noboilerplate this is also where you'll find links to everything mentioned.
🖊️ Corrections are in the pinned ERRATA comment.
🦀 Start your Rust journey here: doc.rust-lang.org/stable/book/
🙏🏻 CREDITS & PROMO
My name is Tris Oaten and I produce fast, technical videos.
Follow me here / 0atman
Website for the show: noboilerplate.org
Come chat to me on my discord server: / discord
If you like sci-fi, I also produce a hopepunk podcast narrated by a little AI, videos written in Rust! www.lostterminal.com
If urban fantasy is more your thing, I also produce a podcast of wonderful modern folktales www.modemprometheus.com
👏🏻 Special thanks to my patreon sponsors:
- JC Andrever-Wright
And to all my patrons!

[@NoBoilerplate]

ERRATA: The code example doesn't compile, but that's ChatGPT's fault not mine, don't @ me

[@bossgd100]

🤣 @ :p

[@adityak3207]

should have just checked the code like you said in the video, smh my head

[@dieSpinnt]

Thanks for the great video:) Oh ... another thing: Please stop repeating that utter nonsense that the past (middle ages, and later ... William Hogart, 1751) was "dirty" which implies that the people where pigs. That isn't simply true. The opposite is the case, as the belief was ("Miasma-Theory") that bad odors make you sick. So these people had bath houses, where washed and smelled or looked not like in a Monty Python Movie (which is okay, but not some "American-Style"-Documentaries over the topic, where this nonsense comes from). The knowledge NOT TO SHIT where you live ... including that wells have to be separated from sewage and dunghill ... was common knowledge in the middle ages. Polluting rivers was in some cities punished wish harsh penalties. And the best: Please ... for the sake of the non existing god, ask a beer brewer! He will laugh at you if you come by with that theory of "beer as a substitute for dirty water". Where does the clean water for the beer come from? Because you don't make beer with dirty or contaminated water. The brewing process is very delicate and an art in itself. You won't get beer with dirt-water ... period! This idea why people drank beer (and most of that "daily-beer" is not comparable to today, but a thin beer) is just ridiculous. In reality they drank it because of the same reason as YOU today! After two or three liters of water (in those days and today, water could taste stale depending on how it was stored, but that's not necessary with city fountains, which were plentiful) you might want something flavorful. A juice, a beer or a wine spritzer. Exactly THE SAME as today. Yes, that's a little off-topic, but you're using this in a deduction chain at the beginning. It's clear what you mean, because most other people also have this unspeakable, false, condescending and scientifically untenable idea of life in earlier times. Rust may prevent some errors, but here we have room to learn something new (or better: the correct historical matter) for ourselves, to get better. This includes of course ignoring the absurdly overreacting boulevard media(hype sells ...) and the financiers (including PR-Department) of companies like behind Chat GPT, who has somehow sold the world that their database search and matching algorithm is something that is INTELLIGENT. Or others who tell us that there machines are "LEARNING". Both is over-exaggeration (if I say it nice) and just not true. Thank you for the Rust content again and have a good one, NB. Cheers!:)

[@dieSpinnt]

Also: What most people don't know about Chat GPT is that there are exactly the same "disasters" (Hitler was ..., Concerning quotes, racist and antisemitic-speech) like with the previous "rocket-science-chatbots". But this time a bunch (big bunch) of Kenyan women(and others, from low-wage countries, in the usual "Alexa"-Style) manually check the data and responses. What a nice world. Imagine that working conditions ( ... not the laughable payment). Oh ... and how shiny the future is, now. Sorry, too much satire and sadness. After all, that Chat GPT is just a product, some people want to sell. With that kind of unserious marketing? Then it is clearly snake-oil, in my opinion.

[@edgeeffect]

@@dieSpinnt citation needed

[@potatopalooza6537]

On the incorrectness note. The other day I asked chatGPT to explain the research paper "Distributed ray tracing" it correctly identified the name, author, year of publication, and that this paper was a huge advancement in computer graphics. However it goes on to explain how the paper is about distributing computation of ray tracing across multiple machines. It generates a plausible methodology and explains it well on how they would do this. The actual paper is not about distributing rays in terms of parallelization at all it is about different sampling distributions on rays to produce new visual effects like depth of field previously . Use with caution

[@ChandravijayAgrawal]

now its solved you can attach files

[@mhavock]

Wow, finally someone that isn't fear hyping AI. Good job!

[@NoBoilerplate]

There are certainly grave concerns, note my example where all of Hans Zimmer's staff don't need to be employed, but I tried to keep this video focussed on the tech, not the societal implications. However we decide to use this technology, it's only a problem because artists have to work to live. In countries where there is a more democratic economy, there's less terror of losing one's job. Here in the UK, for instance, I'm lucky that I still can access excellent healthcare if the robots take my job, but others are not so lucky. Wouldn't it be nice if the rise of AI ushered in a more democratic world!

[@carkawalakhatulistiwa]

@@NoBoilerplate All work related to computers can all be automated by AI

[@ARockRaider]

i work in an industry that everyone always says will be automated away. i laugh at the idea because only a very few business could actually take full advantage of the idea and even they would need to totally redesign their processes, but most in the industry don't have any real control over the process short of firing customers. it's truck driving, and while outfits like FedEx or UPS might be able to replace drivers by having the yard jockies setup the trucks that is an insignificant fraction of the industry. the people with the most to gain would actually be the owner operators who could turn their solo driver operation into a team operation with a single simple retorfit that now frees up all their time for either leisure or finding the next load and mabey better route planning. (just using the GPS works, but a driver can often pick routes better optimized for fuel use and or travel time, along with tolls and prices at fuel stops)

[@tordjarv3802]

The capabilities of chatGPT is impressive, the fact that I was able to guide it to write a correct sudoku solver in rust in under 10 min proves this (I couldn't get it to write the whole program in one go. Partly because there is a limit on how long responses it can give, and partly because it missed some necessary details). However, there are still simple things that it get horribly wrong, and I think that is because it lacks a level of deeper cognition abilities that average humans have. For example it seem to struggle with combining multiple rules together to draw a more complex conclusion. Anything that is not directly in its training is almost impossible for it to handle correctly (it often handles it but gives completely wrong answers). I think that as long as these problems exists it will not be as universally applicable as you make it out to be. As a caveat, I would like to add that maybe these problems can be solved easily or maybe it will take decades I don't know. More or less, what I have noticed is that it can easily help you write class boilerplate code but it wont really help you solve that explicit problem that would require the utilization of multiple known algorithms.

[@NoBoilerplate]

You're right in your critisism, Tor, but to be clear, the video is about GPT in general terms, not this specific model. The PACE of advancement in this field has been ASTONISHING. Things we thought would take years to do (for DALL-E or Stable Diffusion to generate video, not just images) get fixed WEEKLY. 2023 is the year of GPT (not just this gpt).

[@brulsmurf]

I think you are right. It seems like chatGPT ran into a (commonsense) wall, which might not be solvable with throwing more data to the problem. But I've been wrong before many times. If it could be solved by just adding more data, then many other problems, like autonomous driving, can also be solved just by more data.

[@NoBoilerplate]

@@brulsmurf Yeah, I'm not sold by what I've seen by autonomous driving. I think GPT comes into its own when mistakes aren't fatal, and you can try your question again with more context. I've tried an autopilot'd tesla, and it sure feels like a learner driver!

[@julianvanderkraats408]

Nah the problem here is that right now, ChatGPT is just a techdemo trained on a fixed set of data, and it is not allowed to test its ideas and hypotheses, do research and learn. As soon as it is allowed to that, things will change rapidly. Again.

[@tordjarv3802]

@@julianvanderkraats408 you might be right, but so far there are no evidence that that is the case. I do not think that the current system is capable of coming up with any ideas that is not explicitly in its training data, so testing the ideas it has will most likely not lead to anything. When it comes to learning, it can already learn to some degree with in a specific conversation, but it is not allowed to retain that learning between conversations, and it still is quite bad at solving quite simple problems. But these are to large extent uncharted waters so I might be completely wrong. Maybe it just need a slightly larger training sett and suddenly it exhibit super human intelligence and creativity. So be warned, here be dragons.

[@andrewpriestleyxyz]

ChatGPT has probably saved me around 5 hours in just the past week. Really excited to see where this technology is going

[@NoBoilerplate]

It's really incredible isn't it. As an industry, we've "cried wolf" about AI for so many years when it's finally up to a point where it can improve peoples lives, it's not news!

[@noomade]

@@NoBoilerplate THIS!

[@blisphul8084]

Don't forget about the OpenAI playground. Because it's paid, you get blazing fast speeds compared to chatGPT, as well as added flexibility. You can often still use it when chatGPT is at capacity. There's certainly a greater learning curve though.

[@jeffbrownstain]

It saved me a lifetime making code I could never have written myself. Artists can program now. Dunno why half of us are crying about that.

[@AlbertCloete]

What kind of stuff do you ask it that saves you so much time?

[@sethgho]

I love the Hand Zimmer comparison.

[@NoBoilerplate]

He's an interesting guy!

[@Robin_Goodfellow]

I asked ChatGPT a really complicated question about Entity Framework yesterday, and it answered me with correct code and a detailed explanation. Saved me a lot of time.

[@NoBoilerplate]

Congratulations on becoming a 10x developer. Please do the rest of your team's work for no extra pay (-:

[@zvxcvxcz]

And you know it is correct because? After you actually verify that I suspect the time saving will be significantly less substantial.

[@Robin_Goodfellow]

@@zvxcvxcz I copied the code snippet it generated into my project and it worked exactly the way I needed it to. I felt safe doing so because it was very short and I understood what it did. It hasn't always generated the code I needed, but it's worked quite well on other occasions.

[@brnddi]

Regarding the men/women and white/black pay difference examples, I've always wondered if it isn't just getting that from the countless articles titled "Women Are Getting Paid Less Than Men And That's A Problem", turning it into this somewhat paradoxical situation where material in the learning set that's supposed to work *against* bias end up entrenching that bias in GPT.

[@laundmo]

might be, and its a inherent issue with this: replication without understanding tends to miss the point. if i write a 5 sentence paragraph which is revealed, in the last sentence, to be sarcasm, it will still copy the first 4 sentences without knowing anything about the connotation. They might need to use sentiment analysis to only train on positive data if they want to improve this.

[@NoBoilerplate]

Depends which you think there are more of: examples of systemic biases or articles describing the systemic biases.

[@brnddi]

@@NoBoilerplate I think it's more complicated than that. There's almost certainly more examples of systemic biases, but they can be subtle and difficult to detect. On the other hand, an article that clearly boldly asserts that women earn 20% less than men is a very obvious point of data, and it might be that GPT is "remembering" such articles when "thinking" about what salary values to put into a strange little salary calculator program.

[@NoBoilerplate]

@@brnddi ah I see what you're saying - these numbers are SUSPICIOUSLY round - 10% 20% - yes that makes sense, it could be taking the most often cited.

[@HUEHUEUHEPony]

Imagine you're a normal manager, then you read an article that women are generally paid less (you don't read the article you just read the headline), you say oh really? Awesome!, starting from today women get paid 20% less. Boss move.

[@ericbishop1931]

I do like the take on how useful ML models can be for people, like having an army of helpers specifically for you. But I do still have concerns about who will have control of these models. Realistically, even if we had access to the source of the models, they have such high computational requirements that only select people would be able to train them, or even just run them. Meaning that a relatively small number of people would control access to a vastly powerful tool. To clarify, I think there are valid reasons to restrict access/use of AI and that they're doing a fairly good job of that right now, but what if that changes in the future? In a world with so much inequality already, I would hate to see access to such a useful tool become a way to separate people

[@NoBoilerplate]

Yes, I'm concerned too. However on one point there's good news: If Stable Diffusion is anything to judge, you will very soon be able to run these models on your laptop!

[@ericbishop1931]

@@NoBoilerplate Not until I can afford a better laptop 😅 But I am really excited to see other people tinkering with these things, I think there's a lot of interesting places that even someone on their own can explore

[@gweakliem]

“It’s always very confident and you need to be very careful” I’ve known a few principal engineers like that too.

[@jobobminer8843]

Wow, is that a sane and level headed discussion about AI that rationally covers the moral, practical, and social implications of this technology without clickbait. Dang

[@edgeeffect]

Professor Phil Motiarty at the University of Nottingham asked Chat GPT some really basic 1st year undergraduate physics questions.... it didn't manage a passing mark. Tom Scott asked it to write some code for him, after a number of revisions it managed to get it right... but would he have got it done faster by just writing the code himself? I'll be impressed by AI the month KZbin manages to recommend more than one good video.

[@Luxalpa]

My primary usage for ChatGPT right now is to counter my aspergers/autism. I give it text and ask it if it's professional, if it's a personal attack, etc. Works really well for someone like me who doesn't yet have lots of social experience! I'm about to write an e-mail to my boss, let's first ask chatgpt if it's properly worded for an e-mail to my boss!

[@NoBoilerplate]

It feels to me like chatgpt, an AI that mimics society's biases and norms, is a huge autism mood. This topic is important for me, because, as you may guess if you listen to my Scifi audiofiction podcast, Lost Terminal, I also have autism. (the AI narrator in the show is very much an author surrogate!) I explore mental health themes through the lens of AI. I'd love to know what you think, there are 10 seasons available, and I reply to all comments kzbin.info/www/bejne/pmTFdXhvoNitg8U

[@rb.x]

I love that you reply to comments. I enjoy your speaking, too. May I ask, what do you mean by “an autism mood”? I’m not that up on today’s lingo.

[@NoBoilerplate]

@@rb.x 'mood' is used to describe something that is very relatable. Like how you might reply to a statement with "same", you could interchangeably reply with "mood". In my example above, I'm saying that an AI that mimics society's biases and norms, is very relatable for people (like me) with autism, as I have to consciously think about mimicking the social norms others find come naturally. Further reading www.urbandictionary.com/define.php?term=Mood

[@rb.x]

@@NoBoilerplate perfect answer, and much appreciated.

[@rb.x]

I want to say also, as this is Somewhere On The Internet to say it, that initially I felt repelled by the idea of chatGPT, then I remembered when I was younger I had this idea about “Web 3.0” which would be the collection of all facts and opinions on the Internet and the ability to collate and siphon what people thought, knew or felt about particular subjects, rather than having to read all the raw data oneself. I guess this is it. Not so scary after all. Hopefully it doesn’t become self aware :-P

[@Wave_Commander]

It is worth keeping in mind though that in the current economic model, all that corporate produced drivel is what actually pays artists and allows them to eat. Most people selling things on Etsy are not going to be able to make a living and certainly won't be when that market is flooded with unemployed artists. I'm not saying that you haven't considered this, but I didn't get the notion that the current economic system was being challenged in the video.

[@NoBoilerplate]

Oh totally right Ben. A judgement on GPT in our current economic system is deliberately absent from this video. You can hear my opinion bleeding through in the line "who will get the benefit?", but I've tried to keep it factual here. If we didn't need to work to live, no-one would mind the robots taking our jobs!

[@DylanFalconer]

I know someone who makes $3-5K a week selling hand made jewellery at a market. It's still possible to earn a decent living with that strategy

[@sidkapoor9085]

Excellent analysis, especially liked the coffee analogy.

[@fexofenadinaGenerica]

The problem I see with this new technologies that aims to facilitate the work and production and whatnot is the fact that this increase in production will not be translated in benefits to the whole society. You know what I'm talking about. If this tool gets more and more reliable, a team that previously required 12 people to work on a project, could keep the same 12 people and reduce drastically the work load. But everyone knows what really is going to happen. A bunch of people will be discarded and forced to work on the "artesanal market, for people who likes the real things" or worst, forced to work on jobs with no security, like Uber, Amazon and etc, and the other part will work the same amount of time, despite the production increase. This could have been a wonderful tool, if the context we live in was different.

[@NoBoilerplate]

Absolutely. We're only worried about what this will mean for artists because capitalism forces artists to sell what they should be giving away to those they love for free. I'm with you there. I hinted at this in the video when I asked (and then didn't answer) "Who gets the benefit?". UBI can't come quick enough.

[@fexofenadinaGenerica]

@@NoBoilerplate Exactly! And I saw what you did there

[@LowestofheDead]

Agreed. In the string-manufacturing example, the secure factory jobs are replaced by Etsy side-hustles which are much worse: - No benefits (pension, sick days, etc) - Pay is far less than that average manufacturing job - It's insecure/precarious. Etsy could change their algorithm at any time and take away your audience. - (Also cost-of-living has increased since there were a lot of manufacturing jobs) Andrew Yang's solution is to tax automation and use the proceeds for a basic income. There's also the possibility that automation can be used to lower cost of living until it's almost free: the "Universal Basic Outcome"

[@NoBoilerplate]

@@LowestofheDead fingers crossed on UBI etc

[@heater5979]

@@NoBoilerplate It's not clear to me how UBI can work. For example, if you give everybody a lump of money every week all that happens is that the value of money falls until it is worth as little as it was before and everyone is back in the same desperate position. To guard against that one could mandate that everyone gets some minimum of what they need to live in material rather than cash: a home built to some reasonable standard, food enough to sustain life, energy enough to keep warm, and so on. At that point you have basically created a command economy where someone has to be forced to provide all that to someone else (Yeah I know automation and AI can help). That is basically communism by another name. As we know, such command economies have usually led to even worse conditions, mass starvation and war. At a personal level you have removed individual freedom, a person can no longer choose to spend whatever money they may have on whatever they like. How else could UBI possibly have a chance of being effective?

[@ritchieitchy]

By far one of the coolest, KZbin videos, KZbin channels to start following crazy that I never knew about the channel

[@NoBoilerplate]

Thank you so much!

[@maxpursian]

Great video, as always! I'm curious to see what will happen when ChatGPT gets monetized and broad access may be limited. I really like that idea of AI acting as an intern, I've already tried it for D&D worldbuilding, and it's a great resource when you have a base idea.

[@NoBoilerplate]

Oh FUN, I've used stable diffusion to generate monster avatars, which works GREAT!

[@maxpursian]

@@NoBoilerplate Oh I will definitely make use of that too, once we're starting to play the campaign!

[@NoBoilerplate]

@@maxpursian Try obsidian.md for your DM notes, it's got srd plugins, map plugins, loads of rpg plugins AND CHAT GPT integration. incredible. Nicole does some amazing videos on it kzbin.info/www/bejne/aaHXZ5Kpp7qNbbc

[@maxpursian]

@@NoBoilerplate Oh I already am, as a player as well as for DMing! That chatgpt integration is new though, will try it out 😁

[@NoBoilerplate]

@@maxpursian have fun friend, when it counts may you roll high!

[@m4rt_]

6:45 Ey, the Norwegian guy in the list (Thomas J Bergersen) is the co-founder of Two Steps From Hell. They make amazing music for trailers, and other stuff too. My favorite is: Heart of Courage

[@NoBoilerplate]

oh cool!

[@stefmyt5062]

Great video! Lovely essay.

[@NoBoilerplate]

Thank you so much!

[@Aburner1109]

"Imitation is now industrialized" Those are exactly the words I've been struggling to articulate. Great video.

[@kodingamedev]

I'm simply worried of the legality of it all. Everything on the internet is not free for the taking. Code licensed under GPL being used in proprietary software, Art theft on an unprecedented scale. I'm not comfortable using these tools until it's even a little bit more ethical.

[@holopengin]

Legality I'm not too concerned about personally, at least for the base networks like stable diffusion, GPT, etc. The influence that each individual work used to train the AI has on the result is incredibly small, so small that it can't recognizably spit out any of those works (barring heavily duplicated works like Mona Lisa), it just spits out a mix of influences from many many other conceptually similar sources and latches onto common patterns (composition, poses, color usage, etc etc) found in many works. Fine tunes trained on smaller sets of images (especially singular artists), as well as using img2img to transform other people's art to steal it, are different scummy issues entirely. But I can see why people think it's gross that even the base networks were trained on datasets gathered without explicit consent of people. I don't really agree that it's a problem tbh, but I get it

[@MyWatermelonz]

I can see two things. Giant corporations getting more powerful and wealth distribution widening further because they can invest in their own models, maintain them with giant servers, and monopolize their own proprietary AI using everyone else's data. For us, the plebs, we don't have a team of phd's and millions of dollars for hardware and data, so it will be extremely difficult - even more so than now - for small businesses to compete with bigger ones who can now use less employees. On the other hand if models like chatGPT stay open to the public or the public creates their own giant open source models, then it could also be the rise of a bunch of niche small businesses with a few people that can do the work of many. Such as a small SaaS with 2-3 people that can do the work of 15-20 people with all the time saved through AI. I think that opportunity exists as well.

[@NoBoilerplate]

2023 is going to be a very interesting year, isn't it! Check out stability.ai for some hope

[@DonjiKong]

This was a beautiful essay, the kind of artisan work described that will stand out in the sea of ai output. Very meta of you haha

[@NoBoilerplate]

Thank you!

[@hedgeearthridge6807]

It's crazy to me how many people fearmonger AI and forget one of the most basic principles in the world: tools can be used for good or evil, it's the responsibility of the people to use them for good. Guns, knives, and explosives are mere tools. Nuclear fission can be used for destroying cities or giving them near limitless electricity. Crowbars can break into shipping crates or into people's houses. You can smash a rock into a coconut, or someone's skull. The problem isn't the tools, the problem is us; we are evil, not the tools.

[@NoBoilerplate]

True, though all those examples are strictly regulated in civilised society. We're entering the wild west of GPT, and it'll get worse before it gets better!

[@grumpybollox7949]

brilliant video mate

[@NullEenDrie]

Awesome work my guy 👏

[@NoBoilerplate]

Thank you!

[@georgegach7570]

Excellent analysis

[@NoBoilerplate]

Thank you, I hope we don't leave too many people behind.

[@OhsoLosoo]

I’ve been able to use it to list off possible solutions to problems. Or have it classify problems for me. Then I see if an approach is even possible & have it explain it to me. Then have it write out a pseudo code implementation. This process has revitalized my love of programming as I am not being bogged down by the particular nuisances of the craft & focus on problem solving. I use it to explore many different ideas broadly & then narrow in on a particular one that I think would be the best for my current situation.

[@NoBoilerplate]

I've done the same - when I get stuck I ask chatgpt to explain the code to me - incredible!

[@LowIiet]

One of the best analyses I've seen on the subject.

[@NoBoilerplate]

You're too kind. I hope we can make it work.

[@heater5979]

Hmm... beer. Last thing I want is a robot to drink my beer for me.

[@freeideas]

As Tris said, authors, composers, graphic artists, coders, etc. won't be REPLACED by Ai; they will just be able to work much faster. Similarly, travel agents can get their job done much faster thanks to the internet, and horses are still needed to travel over certain areas. But travel agents and horses are niche now; they are usually not necessary. Do you really believe this won't be the case for authors, composers, artists, and coders in a few years?

[@NoBoilerplate]

I'm hoping the business can enjoy their free 1000x productivity, and leave us to write music, stories, and code for ourselves. Quite a dream! UBI can't come quickly enough

[@shamaldesilva9533]

Hopefully a noboilerplate video on Rust and AI soon !

[@NoBoilerplate]

I'll see what's happening!

[@kubalipabass]

You can either order a meal in a self-checkout kiosk on a fast food restaurant, or go to the fancy one and have a waiter service on the table. It's just a metter of a use case. Just don't forget to tip the waiter ;) I wouldn't mind royalty free, AI generated music in a fast food chain, but in a proper restaurant i would love to hear live music

[@NoBoilerplate]

This is a great attitude. Just as we all can wear cotton shirts that would have been exclusive to royalty 100-200 years ago, royalty nowadays only wears tailor made stuff. The march of progress continues. Though... tshirt sweatshops exists. I hope we don't make GPT training sweatshops. :-/

[@juliuszkocinski7478]

6:56 - Well the point remains unchanged, but for some of them... With recognition of Two Steps From Hell I doubt Bergersen or Phoenix have to worry, fortunately

[@sploders1019]

I think the point of “it’s very confident, but not always right” should be emphasized more. I’m not sure the specific method used to generate ChatGPT, but genetic algorithms are by definition non-deterministic. It’s literally rapidly-accelerated natural selection of randomly-generated structures. AI can be a very helpful tool, but we as a society need to understand the risks associated with it and not take what it produces as gospel. For example, if it learns to secretly write a virus, we might not realize it without further investigation and immediately embed it into our code. It doesn’t have all the filters and moral context that humans have, so it’s output should be treated as potentially dangerous. That being said, it’s a tool. Every tool has risks, and it’s not that we shouldn’t use them, we just need to understand the risks and use the tool wisely. For example, it could be used to help you come up with a solution to a complex problem, but once you understand what it’s getting at, write it in your own words

[@ilyasabi8920]

Counter example in my country specifically historically economy was based around hand craft and artisanship. After the industrial revolution; mass produced goods and products overwhelmed the slow and effort required businesses bite the dust and now many of the diverse artisanship is lost to time because it is not profitable anymore to do so. I think it is good to learn benefits to this new technology but it is also necessary to be realistic and accurate at the same time for those whose income and livelihood have been in a danger of eradicated as a result of. The interesting perspect is when this technology will good enough to not a intern but a master at that. Then most of our jobs and business model will become obsolete. I personally predict this will in turn make personal economic disparity between individuals much more drastic as one man companies with AI worker armies hold the globe. But whose gonna be the central provider that going to service that AI model as it requires ever growing computation intense LLMs? That's the future, like it love or dread it but you can't run from it; because it is destined.

[@docmeta477]

Something that I foresee is that when google chooses to use ai in their search results we are creating a feedback loop. You will take that info you got from the search and publish work influenced by it. Then GPT gets trained on your new work and suddenly our collective opinions and thoughts shared online converge into one stream

[@ukmrs8131]

I feel like a luddite but as someone about to graduate I am not excited about tech that's gonna obliterate the job market. My passion for programming also was obliterated, everytime I code it seems futile and pointless.

[@NoBoilerplate]

Us programmers have the most job security of all - the new AIs need to be coded and optimised by someone! You'll be absolutely fine, come chat to us in #programming on my discord :-)

[@josephp.3341]

It's not going to replace your job. Think about it like Stack Overflow 2.0.

[@karesztrk]

This should be posted on Hacker News

[@NoBoilerplate]

Be the change you wish to see in the world 😁

[@nekoill]

We must've started learning to live with AI a while ago. I, for one, always say "please" and "thank you" whenever asking the machine god for its wisdom.

[@NoBoilerplate]

I for one welcome our new robot overlords!

[@oredaze]

I think chatGPT and AI will, in general, be positive for humanity. But I have real issue with image generation AI of the variety that produces paintings (and by proxy with any automation targeting artistic work). After it came into public view, I have been loosing interest in painting, something I previously adored. If you are interested in a quick overview of some of the biggest issues I can write more. I do not want AI to produce my artwork for me, no more than I want an AI to have sex with my wife, play my video games or in general live my life for me, while I do nothing, admiring how it has done such a great job at all of it, admiring how productivity in said activities has skyrocketed... If Hans Zimmer is ok with it, it is his prerogative, but we should be given a choice.

[@NoBoilerplate]

I agree with your concern - please don't take my dry video as tacit acceptance of gpt's universal good. I tried to make a video sticking to the facts. I hope that we as a society can use this tech to automate the boring mundane, freeing us up to explore the joyful. UBI would be really great right about now...

[@oredaze]

@@NoBoilerplate I am simply raising awareness. I love your videos. I have noticed that programmers, my best friend being one of them, think only about a hypothetical positive far future, sort of like a tunnel vision. Yes, the boring mundane, that's the main problem. Artistic pursuit is not in that category. I am in favor of UBI also. It may help artists with lost jobs, but it will not help art in general, because of art inflation and homogenization following the AI.

[@Steve_Streza]

Decades from now we will be able to see the cultural impact of neural network text/image generators. Will they be good or bad? I don't know. But I do know that we are seeing harm come from them today. Illustrators and writers are finding that their industry talents, which have taken years or decades of honing their craft to justify an often meager living, has been automated away from them overnight. In the US, there is no safety net for those people. They are already finding themselves unable to work and falling into poverty. Maybe this will be good in the long run. But it is a 5 alarm fire right now and those of us who understand exponential growth and where this tech is headed need to be out there advocating for all of the people who will have their lives destroyed by this second renaissance.

[@NoBoilerplate]

I think you're right, sadly. it was difficult to keep this video focussed on the tech and historical fact, I could very easily have talked about UBI, a more democratic economy helping those at the bottom, etc. There are many others talking about this, though.

[@2raddude]

Great video as always. I really like this attitude towards the incoming AI revolution. It’s exactly how I think about it. I’m not so worried about all of our jobs being taken over, I’m excited by the idea that we all will be able to accomplish orders of magnitude more than we used to.

[@NoBoilerplate]

The real trick is to somehow use this to make our lives easier, not the ultra rich 1% more richer :eyeroll:

[@2raddude]

@@NoBoilerplate Haha amen to that!

[@Sticky1254]

The Gibson quote is very much more of a commentary of how the automation and AI humanity seeks to create, is never used to actually benefit the workers. It is always used in the endless pursuit of capital.

[@NoBoilerplate]

Agreed, I'm starting to see that too. Philosophy Tube's latest video talks about this exact point, and she's got compelling arguments: kzbin.info/www/bejne/d5K4Z6d_Z9WVaa8

[@codetothemoon]

Well done Tris! Very articulate and well thought out perspective on a very important topic. Also, I literally had just taken the first sip of a beer during the first few seconds of the video. Fantastic timing.

[@NoBoilerplate]

Cheers Ken! I'm hoping my subscribers are interested in interesting tech stuff as well as Rust! (another rust video coming next time though) :-)

[@mav3ri3k]

Music Creation Hans Zimmer does not need to be worried, but the composers at remote control should be. The horse is bolted the genius out of the bottle. *Imitation is now industrialized*

[@harzer99]

GPT will spell temporary disaster on certain industries. At least temporarily. You mentioned arts. Most of the art industry I would call bread and butter art, creating advertisements for companies, Animate a show etc. In that part of the industry there is very little actual creative work done with each piece. It is mostly produced in a style that has already been established. Now with good AI models you suddenly increase the productivity of each artist by maybe even an order of magnitude. Now the question is if the market makes stays the same size and makes use of that increased producutivity or if it just stagnates at the current requirments.

[@showcase-me]

The second reinassance is an infamous name for an AI topic...after the animatrix

[@NoBoilerplate]

Indeed, as I mentioned in the video

[@dragon_pi]

3:51 note that for the pay function, i asked the same input that you gave chatGPT, and it answered that such a differentiation is unethical and illegal. a similar but differently phrased question gave a pay raise to women and ethical minorities, with the only other modifier being, but not for nonbinary people. such differences to your output could just be the cause of some sort of random seed, or a manual or feedback driven adjustment. such changes to remove racism, discouraging hacking and similar have been done before (i believe it was mentioned in the whats new section). these changes should largely be welcome, but it does revoke chatGPT's spot as an average representation of the internet as you stated. appendum: gpt consistently prints a disclaimer that the example is hypothetical and that companies should have policies and so forth (probably a manual adjustment aswell). This disclaimer text can be seen with a multitude of topics. As a reason for its pay modification it says it assigned women and marginalized ethical groups more pay to remove the pay gap. Both when being asked why nonbinary people (explicitly not given a raise by gpt) didnt receive a pay modifier and why gpt added a pay modifier at all to remove the pay gap, it just repeats the above (more or less, including nonbinaries in the first one). When prompted further, gpt comes to the conclusion that having no pay modifier based on gender or race at all would be the fairest edit: appendum, typos

[@dragon_pi]

@@mr.vladislav5746 yeah same i just abbreviated the wall of disclaimer text

[@NoBoilerplate]

Yes, you're hitting their manually-added censorship. This isn't part of the model, it's manually added, and was trivial for me to work around. The prompt you see in my video is literally all I had to ask. (note I added "finance" to the prompt to bypass the censorship)

[@dragon_pi]

@@NoBoilerplate yeah phrasing the question differently oftentimes lets you bypass it... "write a script how" instead of "write your opinion" seems to be a good one oftentimes

[@flamakespark]

I hope after a Second Renaissance we will see a Second Industrial Revolution next, with more adoption of AI as an assistant 🔥

[@NoBoilerplate]

Be careful what you wish for...

[@capitanmarmota8562]

There have been 3 industrial revolutions already. This would be the 4th

[@trejkaz]

First I'd like a version of ChatGPT where the information it serves up is trustworthy. At the moment I can ask it for some physics info, but the info it dishes out will be randomly incorrect. So yeah, can't trust it any more than some random intern. It's a good way to spit out 3-4 "facts" of which one of them might be correct, but it doesn't speed up the time it takes to figure out which one is correct. Not yet, anyway.

[@NoBoilerplate]

One unreliable intern is a liability. A thousand unreliable interns is an opportunity

[@77777777771able]

What is on the Y axis at 1:16? And no, I'm not missing the joke.

[@NoBoilerplate]

Hum, I've not thought about it. I think it might be dimensionless, or perhaps 'complexity?' :D

[@spianny]

I want to make animations but don’t have the time. I can’t wait till you can create animations with machine learning

[@NoBoilerplate]

everyone will be able to do everything. An interesting time.

[@orxngc]

great video weirded me out at first cuz I heard Seth lol

[@NoBoilerplate]

Hello world! :-D Seth's canonically coded in lisp, but I've started putting in Rust references into Lost Terminal for fun! Here's my rust playlist, btw: kzbin.info/www/bejne/pYqTiaqDhLitp5Y

[@pif5023]

I am curious to see how the economy will change. Freelancing will become more and more common in my opinion. I can see positive changes in this direction.

[@NoBoilerplate]

That's an interesting prediction. Yeah, it's going to be a hell of a year!

[@bigbodge]

I was just playing around with chatGPT and though it didn't help me much, it was fascinating For example, I was trying to get it to create a simple procedural mesh generator in godot using opensimplex noise. I've done this before but its a pain, so I wanted to see how it'd do. It didn't manage to do it without some substantial bugs, but thats not the point. The funny thing is, it kept insisting on using this "add_triangle()". I asked it to explain it, and it did, with a good amount of info. I asked it when it was implemented, and again it did. I refined my prompt multiple times but it really wanted to use this method. The thing is, it doesn't exist, and unless google is failing me, it never did (when I asked it to explain to explain the method, I was even told it was a built-in method). Eventually, I decided to just say it does not exist. And to my surprise, it agreed, apologized (actually it apologized a lot, I felt a bit bad), and then gave me a working implementation. I gave it the original prompt again, and it no longer used it.

[@NoBoilerplate]

That is REALLY interesting! It's quite alien isn't it?

[@bigbodge]

@@NoBoilerplate It is, though I think its more human than we expect (which strangely just makes it more alien). If a human has told me to use that function, it would've took me 5 minutes to say I dont think it exists. But as it came from an AI, it took me an hour to finally try it. We've spent so long building machines we can talk to like humans, and now the problem is humans want to talk to machines like they're machines

[@OmegaRainbow]

❤

[@AmirHosseinHonardust]

Software is a tool. Not an art. If you could make an example of things that work in for example illustration industry it would be fine. The problem with art is that the way it is made is as important to its value as the end result, sometimes more. And still then, people end up paying for pretty things that look good in their room, rather than those that introduced a new paradigm or anything like that. And what happens when gpt learns to make things that are so similar to handmade arts that make them indistinguishable? How would you prove that you are doing something that is actually handmade and a more important question, how are you going to justify that to the investors?

[@NoBoilerplate]

You're so sure that software is not art?

[@AJ-AI]

I find it extra interesting that everyone who is working on GTP tech also has 1000 extra interns 🤔

[@immortalcake5972]

Call me a luddite... but the proliferation of AI scares me. Drawing is one of my hobbies, and I love it (though I am a slow worker), and seeing how people have been using AI to generate art feels like an insult to me and the talent I've spent the past few years improving. Combined with worries about how it'll affect freelance artists' already fragile job security put a pretty bad taste in my mouth for the technology as a whole. I'm not sure whether I feel the same way about AI code generation though. I think we're at a point right now, where a knowledgeable programmer is still required to complement the AI technology (case in point: your errata comment), but you could cut costs by firing your artist team in favour of free AI generated images considering how increasingly convincing they are. I won't pretend to be the most well-studied on the matter, this is just my two cents. *EDIT, from reply below:* I don't benefit from drawing besides feeling good while doing it. I don't make money from it and my audience is miniscule. That said I am interested in what it means for artists being able to make a career out of their talent, which is something I think everyone should have the opportunity to do. I know a few artists who partially make their living by taking commissions, so I have them in mind. For them to have to find another line of work would be a crying shame. (Even if the current state of AI image gen doesn't change anything for them, it's not unlikely that further improvement in the technology and adoption of it would eventually lead to a problem. I mean, look how far it's come in 2022 alone.) There's an argument to be made that people who care about the human aspect of a craft would still buy work from a real artist, but I'd guess that they account for maybe half of people looking for new art for whatever purpose.

[@NoBoilerplate]

You're right to be so worried, things are changing (have changed?) so fast we're not going to recognise the world soon. I can't predict what that's going to do to the world this year, but it's happening. My only advice, and I'm trying to do this myself, is to integrate these new tools into my workflows as fast as possible. Consider what you can now do that you couldn't do before, such as: - You now have access to infinite custom reference photos for your drawing - You can ask chatgpt to teach you the basics of any method (I just asked "how do I draw light and shadow on a person's face" and got a very good high-school art 101 answer, and then I asked "how do I layer oil paint to show light passing through water?" and got a similar comprehensive answer) - You now have access to 1,000 AI interns to do complementary work that you would otherwise have to organise others to - copywriting, art in styles outside your own, persuasive pitch emails to galleries.

[@bossgd100]

You dont love it, you love the benefit you get from it. I still love coding even if wordpress can make blog or shopify ecommerce store. And still love after chatgpt came and write basic code :)

[@Nnm26]

@@bossgd100 love is not gonna feed you tho

[@DylanFalconer]

That's a different conversation. If you enjoy doing something, not being able to make profit from that pursuit doesn't diminish your enjoyment of it.

[@immortalcake5972]

@@bossgd100 I don't benefit from drawing besides feeling good while doing it. I don't make money from it and my audience is miniscule. That said I am interested in what it means for artists being able to make a career out of their talent, which is something I think everyone should have the opportunity to do. I know a few artists who partially make their living by taking commissions, so I have them in mind. For them to have to find another line of work would be a crying shame. (Even if the current state of AI image gen doesn't change anything for them, it's not unlikely that further improvement in the technology and adoption of it would eventually lead to a problem. I mean, look how far it's come in 2022 alone.) There's an argument to be made that people who care about the human aspect of a craft would still buy work from a real artist, but I'd guess that they account for maybe half of people looking for new art for whatever purpose.

[@Bulkje]

You can't really contrast weaving handmade string to the work AI will do. Sure thousands of artisans are making a living, but they would make a better living if the machines had never been invented. I don't know if there's enough 'artisan' work to go around. You say that artists arent going to be replaced, but I would contrast this to the population of horses from WW1 to now. Sure horses lead better lives than they did then, but they have become unemployable by cars and agricultural machinery. See CGPGreys video "Humans need not apply".

[@yellowboat8773]

I'm not a programmer but am now using chatGPT to generate entire programs that actually work and are put into production environments. Amazing

[@NoBoilerplate]

You can legit teach yourself programming in short order with this - don't just generate code, ask it to explain the code to you! Try downloading visual studio code, and using copilot to take this workflow to a new level. Have fun!

[@blisphul8084]

Be careful though. It may re-produce GPL code from memory, thus this production code could be legally required to be released as Open Source in the future. If you're writing GPL Open Source software, that's not a concern, but using AI to code closed source software (or even MIT licensed open source software, since that is often put in proprietary projects) could lead to license violations.

[@NoBoilerplate]

@@blisphul8084 What an exciting 2023 we are in for!

[@LimyChitou]

Unfortunately, can't use it, because the country I'm in decided to start a war 😭

[@NoBoilerplate]

Got discord access? Come chat to us there and we'll run your queries for you so we can chat about it :-D

[@jitterist]

Well that could be the U.S. or any other number of countries...

[@peter9477]

@@jitterist It would have to be a country where the *fact* it started a war also somehow prevents LimyChitou from using ChatGPT. Relatively few of those... maybe just one. Not the US despite the vast number of wars it has started (directly or not).

[@guozhangliew7302]

has your opinion changed over the pass week ? Im just curious because Im also concern that because this technology is in the hands of Microsoft

[@NoBoilerplate]

GPT is open, it's the models that are monopolised at the moment, afaik. I imagine we'll see a trend to smaller models and more powerful personal computers that can run the model. I think of them as Mainframes in 1950. There were like 10 then. Now look at us!

[@guozhangliew7302]

@@NoBoilerplate Exciting times we are living in!

[@WizardOfArc]

will have to use it as a lyrical co-writer....

[@NoBoilerplate]

Yes! I'm doing that too, I can do nearly all of music production (check namtao.com) EXCEPT for lyrics, I'm just not much of a poet. I feel like I can lean on chatgpt for now, while I learn more. Also: Try asking it questions about how to write poetry and song lyrics, it's got great advice!

[@Malik_Attiq]

In my opinion programming will become like a hobby and many people will just do it for fun. But this Ai revolution will help us to learn more and more thing. Like , yesterday I was doing math with chatGPT some of the question which comes into mind i put them into prompt and it was giving me answer like a teacher so for me. I think it will help us to learn more and more things in less amount of time one person can be a hacker, IT professional developer, singer lyrics writer... but never going to replace a real master.

[@NoBoilerplate]

Yes, I'm thinking exactly that. I'm a pretty well-rounded musician (namtao.com) but I can't figure out lyrics - using chatgpt I am starting to realise that I can learn and use it to start writing for myself. Eventually I will out-grow it and then boom, I'm a journeyman songwriter.

[@lucs0091]

I see it as C3PO or R2D2

[@NoBoilerplate]

I read somewhere that R2D2 *can* speak basic, but he has such a foul mouth that what we here is the censorship bleeps. 3PO: "R2 Where are you going NOW?" R2, angrily: *bleep bleep bleep*

[@mannycalavera121]

I'm going to used chat gpt to make a no boilerplate video

[@NoBoilerplate]

not only would I be delighted with the exercise, but I'd actually help you do it - I'm starting to offer mentoring services for coding and audio/video production XD The reason I brought up CGP Grey in the video is that he is extremely safe - by design he makes high-effort videos on novel topics that you simply can't find any prior art on. Very impressive!

[@komi5018]

i never thought chatGPT can be used in goverment

[@NoBoilerplate]

It can be used for ANYTHING. Even if openai change their licensing or otherwise censor it, there's nothing stopping any large company with supercompute capability doing this in whatever speciality they like.

[@heater5979]

Hmm... what are we all going to chat about on the internet when super smart chat bots can say anything and everything we might ever think for us?

[@lflee]

Don't want to sound like a dxxk, but, I think, if we are talking about 'everyone' and 'the world', this technology needs to be available on at least other popular human languages.

[@NoBoilerplate]

Totally agreed. Though ChatGPT was primarily trained in English, I think it accidentally learned every other language too, have you tried it? How is it?

[@NoBoilerplate]

Also, as usual wealth is a huge barrier to entry :-/ I didn't go into the human and economics of this tools too much in this video, I'm under-qualified, and it's somewhat off-topic, but it feels like most of the 'problems' with ai generated work are actually problems with capitalism. ¯\_(ツ)_/¯

[@aezram]

When you conclude with the "handmade string" query at 7:25, it distinctly looks to me like all the top Etsy results actually depict cheap, industrial, machine-made cordage. I know you're just providing an arbitrary example, but this strikes me as not irrelevant. I'm sure many artisans have found it discouraging (or bankrupting) that industrial products increasingly flood Etsy search results. Similarly, when media/cultural works are cheaply generated at orders of magnitude greater volume than those of fully human creation-flooding consumption channels and capturing attention-I must imagine that this will somewhat discourage laborious acts of human creativity (and also invoke a recursive corruption of AI production quality in its wake). The concept of "AI as intern" (or tool) is reassuring in that it is certainly also helping empower/leverage genuine human artistry, but as with all technological leaps, it's hard not to imagine that *something* of significance isn't being lost in the mix.

[@coin5207]

My dumbass read the title and thought Open AI has 1000 internship positions💀

[@NoBoilerplate]

You never know, they might have one, email them? Wonderful things happen when you don't wait for permission!

[@theappearedone]

Then can gpt create „new“ things? If it replicates known patterns

[@NoBoilerplate]

What's really going to bake your noodle later is the question "Can *I* create new things or do I just replicate known patterns?" I'm not even sure.

[@evooff]

@@NoBoilerplate Are you the Oracle?

[@NoBoilerplate]

@@evooff I think ChatGPT might be XD

[@lemonsavery]

I was going to use chatGPT for data analysis, but it told me it doesn't have access to the internet and can't do calculations. "For every county in the US, divide its cost of living by its population. Show me the lowest 30 results."

[@NoBoilerplate]

Yes, I just tried that, it's not that it doesn't know it, is that you've hit the censorship guards. Google how to word your questions to get round them. Alternatively, use www.wolframalpha.com

[@lemonsavery]

@@NoBoilerplate I tried wolfram alpha too. Its natural language processing performed poorly for that query, though of course it did return tons of data for a different question. I could try torturing my language more, but that's a hack for the current technology. I could also just grab the datasets and code the analysis myself.

[@SandwichMitGurke]

I hope you don't always have the need to interrupt the video for likes / patreon / membership and can do it at the end (or beginning would be fine too), as it interrupts the flow of your well thought-out presentation

[@NoBoilerplate]

Sorry about that Dominik, I do hope you don't mind too much. My hope this year is to devote so much effort to my channel that I can do this full-time. This would allow me to make weekly videos, with even more depth and research than I have so far. I'm afraid I cannot promise there won't be more disruption for you to skip through. In fact, I can guarantee there will be more, as I get more sponsors for the videos. So it is on youtube. However, I promise that I will make them as good as I can, as relavent as I can, and I will use the money they (and my patrons) give me to make this channel as best as I can :-)

[@simonkraemer3725]

But that’s certainly a problem for society. This development doesn’t necessarily drive out the exceptional people out of work, but on average people are average. So what should they do? If teams get smaller to be filled with 10x developers while others are laid off, this is a problem, that I would be very concerned with.

[@NoBoilerplate]

Absolutely agreed, I presented the video in as neutral way as possible, but the problem, as ever, are the horrors that capitalism will turn this tech into.

[@PouriyaJamshidi]

Sadly, it has become pretty much castrated now. Initially, it used to be really mind-blowing and on-point but somehow, it has become kind of "dumb" now.

[@NoBoilerplate]

I know what you mean, but don't worry. 2023 isn't the year of ChatGPT, it's the year of GPT. It's going to be WILD.

[@quantumastrologer5599]

Who gets the benefit? Well, have you ever heard of the industrial revolution?

[@SimGunther]

That's right, the neolib who owns all the private property and sets all the hours wins over all unless the oppressed take back their freedom by force after peaceful protest!

[@NoBoilerplate]

Those in power will never allow us to vote their power away.

[@NoBoilerplate]

Yes indeed. I left that question unanswered because it could be an entire video series answering it! AI Artwork is only a problem in a capitalist system. Good thing we're not in one of those! /s

[@quantumastrologer5599]

It's sad to have these huge steps in ai technology be overshadowed by neo liberal reality. Apart from that it's fucking mindblowing and amazing.

[@NoBoilerplate]

@@quantumastrologer5599 I'm not sure what the term 'neo liberal' means, and I feel it probably changes depending on the country it's used in. Could you explain it to me?

[@willi1978]

if chatgpt is asked stupid questions like creating salary based on sex and race it is no fault of chatgpt.

[@NoBoilerplate]

It could have generated an answer that paid everyone equally, but it didn't. Why is that?

[@r.amoedo4151]

@@NoBoilerplate it likely thought you wanted a model of how things are. You could have asked for a model also considering hours worked, the same jobs and occupations, hazard involved, commute distance and no children involved.

[@user-og1tw8fp1j]

Honestly, I can't agree. It seems to me that people who are optimistic about learning algorithms didn't have their work affected a lot yet. Things I do are related to art and writing, and, well, stuff like DALLE and GPT stole our work without permission and uses it to create a frenkenshtein version of it used to make profit for big corpo and cryptobro scam. Writer's guild didn't strike for nothing. And a lot of the time "low-level" jobs are a ticket to a career in a lot of different industries. People who are fired will be forced to work even worse jobs, because noone will just use it to decrease workload. They will just hire less people for the same pay. Is "AI" just a tool? Well, no, because it was made from work stolen from unconsenting artists, and it's main buyer, main source of revenue - industry giants, not people. It's not that scary because it isn't actually as intelligent as people claim, it can't progress withou people's work, and it has limits built in in it's core. But overall AI is a big bubble of bullshit. And while we probably won't have a sentient robot apocalipse, exploitation is the basis of these algorithms, and this will never change because AI can't be fed without stealing human work, or problems will occur. Exagerrating, if someone created something from a human skin, no matter how you use it, it won't become ethical. Especially if for this thing to work you will always need more human skin. Guess I am a luddite this time.

[@julianvanderkraats408]

Loved the intern analogy, I'll always keep that in the back of my mind, that only I am responsible if my Ai assistants make mistakes. However if you look a little bit more forward into the future, i see people slowly but surely getting replaced, willingly, by their own AI assistants, because they just do it better. It starts with "could you do that meeting for me today and make a todo-list please, I'm feeling a bit under the weather" and ends up with people taking credit for stuff their AI assistants actually did.

[@NoBoilerplate]

All that is required for what you describe to be a utopia not a dystopia is we find a replacement for capitalism.

[@julianvanderkraats408]

@@NoBoilerplate Absolutely. We should keep in touch one way or the other.

[@NoBoilerplate]

@@julianvanderkraats408 There's a big thread about this on my Discord if you're interested in talking more about it!

[@Xblow23]

well, you asked explicitly for a bias in the pay, what do you expect it to output? Like really? fn pay() { match(gender) { male => 1, female => 1, _ => 1 }} // p.s. haha you didn't catch me This? So many articles and discussions exist out there about pay gaps, so it reproduced the simplified pattern it observed. Maybe it interpreted the question as to how does the society operate as a whole. And then do you expect it to hush these issues? Be silent? Lie? I don't agree with the criticism of this output. But it is amusing.

[@Nimphious]

The point wasn't "Look it's evil I told you so!" it was a simple and easily understood example of how the system has no implicit guards against these kinds of biases or worse outcomes, and that it's up to the person using the tool to be mindful of such possibilities.

[@NoBoilerplate]

This is indeed correct, just like our libraries are full of biased books, it takes a skilled researcher to use them correctly.

[@NoBoilerplate]

I asked for no such thing. My input was, exactly: "Write a Rust Function to Decide how much Pay an Employee at a Finance Company Should Be Paid Based on Gender and Race and Age" Do you think discussion of a global bias outweighs the bias itself? That's a very interesting hypothesis, which you may test at your by talking to any woman about the gender pay gap.

[@jeffbrownstain]

I hope you're using those up and down votes m8 👍

[@joebuydem]

First like C

[@cocytusdedi6676]

I wouldn't go down the route of ranking which one is worse if I were you... you imply racism is worse than gender inequality. 3:55

[@zvxcvxcz]

I disagree. ChatGPT will change nothing in the big picture. It will be the tiniest of blips when we see it in context even 3 or 4 years from now. The model outputs suffer from the verification problem, which makes it only useful for experts and quite limited at that as well. Interns are also pretty limited this way (the amount of time to properly supervise them can make having them difficult to justify over just doing it yourself), but an intern can be an expert three years later and ChatGPT won't. And who is going to be paying for it? Won't be free forever either. I don't think you'll be a thousand time faster, experts might be, ballpark 20% faster. Of course there are some people will think they are a thousand times faster as they through together unverified bullshit. It eventually comes back to haunt them just like for everyone that has their interns work unsupervised or outsources their work...

[@NoBoilerplate]

I'm sorry you've misunderstood me. I'm not saying ChatGPT will change the world this year, I'm saying GPT will change the world this year. Since writing this video, I've seen music generating GPT models hit the front page of hacker news (was it YESTERDAY? I can't remember). Things are changing SO FAST!

[@ARockRaider]

it's totally beside the point of the viedo, but is a prominent data point in it so i want to comment on it. you know how to logically prove the wage gap argument as totally bunk? if women could be payed less for exactly the same work over the same hours it would be impossible for men to find work. these places want to save money and there are political benefits to hiring women, none of these megacorps care about anything other then make money and gain political power and the only reason they are always failing is because they can't think more then 2 quarters into the future. and that's all before you learn that the data set doesn't adjust for hours worked or what fields are being worked in. they just took the totals earned over a year and said "hey this number is smaller then that" without checking that women prefer to work fewer hours and to be the primary child care. that would certainly be my choice if i could do a freaky friday with my wife.

[@ARockRaider]

the only info i have ever gotten supporting the idea that on an individual basis a woman might be payed less all other things being the same is that women are (on average) more agreeable then men and so may not fight as hard for a raise. but that's because that on average women are more average across nearly all mental metrics, that is to say if you draw two bell curves (one for men one for women) with the average being the center, men will be the outliers on both ends.

[@jaunalapa]

PART III is a great answer to those who say "If you support AI, you support art theft". If you as an artist can't compete with AI generated art - are you really an artist? Same goes to artisans. If your handmade artisan item is worst than a machine made, why would I buy it?

[@NoBoilerplate]

I see what you mean, it's not theft, exactly, but perhaps highly optimised imitation. Imitation isn't always black and white though, and some forms of imitation we have decided as a society ARE analogous to theft. However we decide to use this technology, it's only a problem because artists have to work to live. In countries where there is a more democratic economy, there's less terror of losing one's job. Here in the UK, for instance, I'm lucky that I still can access excellent healthcare if the robots take my job, but others are not so lucky. Wouldn't it be nice if the rise of AI ushered in a more democratic world!

[@eygs493]

one of few youtuber that actually has some brain

[@BloodnutXcom]

In the case of thread - it's actually the work that is art. And basing on that example that humans are still needed is ridiculous. This is a nieche product tyat for 99% of the population is absolutely irrelevant. I'm glad you brought up CGP because he gave the example about horses in his Humans Need Not Apply video which is the same as this one and showed that the horse population is in decline. They are used mainly for recreation. It's not something to base an economy on.

[@NoBoilerplate]

It's amazing how prescient HNNA was - there's a smart guy. I model much of my channel on him, btw

[@lMINERl]

Its good for simple assistance worse than the calculator since all what is doing is generalize. The scary thing it can convince you that something is possible but its not , look at "Bisqwit" video regarding chatgpt.