It's not that "we built creative AI first" - it's that the most vulnerable, unprotected intellectual property on the open internet just happens to be the output of creative artists without the deep pockets or legal experience to defend themselves. If America's corporations put all their customer service call center data on the open internet (because they are idiots), AI would have been potentially trained on that first - and AI would be amazing at all sorts of customer service. Or if America's banks (violating all manner of laws) put America's banking information on the open internet - then AI might have been trained on that first. AI, so far, doesn't just "happen to be good at" creative things. Quite the opposite. Arguably it's terrible at being creative relative to more simple human pursuits. It's just that artistic activity is highly exploitable under the current legal regime.

@bennycarter5249

14 күн бұрын

Highly expolitable but NEVER replaceable

@LooseOrangeJuice

13 күн бұрын

That's true, although I think what he was saying is that LLMs can help humans become more creative, and I can attest to that being true.

@y1729

11 күн бұрын

He was contrasting creative vs logical AI. The AI we have can’t do basic arithmetic consistently because of its underlying architecture, basically it is too creative!

@LooseOrangeJuice

11 күн бұрын

@@y1729 I believe that was a tangential point he was making. I think he's also saying that even if the AI were more accurate on those types of tasks, the primary benefits would still come from unlocking creative potential in the *human* mind. I'm not 100% bought into that idea, but I'm at least hopeful because that sounds like a better future than humans simply being more efficient at doing the same types of repetitive and unfulfilling work that many people today endure.

@luckwhisker

11 күн бұрын

@@y1729 Since when does being bad at math equate to creativity? Most of what AI produces on the creative front is more frightening, boring or off-putting than anything. The fact is AI as it is right now is bad at almost everything. Only people who aren't creative think otherwise, because they don't know what good creativity even looks like.

It is very clear that IF you can train broad range classifier on good real practical production data that you CAN automate manufacturing and some service tasks which previously couldn't be automated. Importantly, the data has to be good and the process has to be static unless you want to retrain your classifiers. AI isn't primarily a technology for the consumer. It's a technology for producers to multiply their capabilities, but it requires extensive upfront investment.

@NateBJones

9 күн бұрын

That's a great frame!

I think there could be a management issue with the productivity impact of AI. If your workforce productivity doubles, it doesn't necessarily mean you know what to do with those newfound work hours. There also could be something akin to diminishing returns for most companies in the short-run when this kind of productivity shock happens.

@postblitz

13 күн бұрын

You would know what to do if you were a real manager and not some petty bossman hired to curtail the more uppity hires of your coding gulag. Namely: productivity in IT evolves into products if allowed leeway and support from management. This currently rarely happens and very often the employees need to leave, make their own company and plug their work in there to get credence and publishing.

@NateBJones

9 күн бұрын

I think this is an under-made point: management focused on cost optimization isn't necessarily positioned to successfully imagine what to do with a productivity breakthrough

Palantir!📈🚀🔥⚡️

Great points.

@NateBJones

9 күн бұрын

thanks!

To extend the Uber example, this current state in AI is more like if Uber had promised investors that their return on investment will come from a breakthrough in self-driving car technology someday which of course never came. AI investment seems driven by the hype in what LLMs will be able to do one day but the cool chatbots we already have may be the final product with incremental improvements from here.

@NateBJones

9 күн бұрын

That's an interesting take, particularly as Waze breaks out self-driving services IRL in a few cities

It does seem as Altman etc have said that customer service seems to be where LLM efficiency leads most obviously to job loss perhaps… so the most competent email rep can now replace themselves and two other people if they have a significant database of past queries and the product/service is fairly stable - and maybe less voice/carpal tunnel or whatever else stress for that remaining employee IF careful thought is put into everything - so that’s probably not quite enough by itself to justify the money mania BUT obviously that’s only one use case yeah it obviously is already changing many, many things and if LLMs+whatever else make advertising even better however defined in particular then wow that’s what the Wall Street World thrives on indeed yeah more consuming🤖🤯

I think there is a market for AI to replace experts that are dying off. Think Cobol, or niche degrees that aren't around or popular anymore.

@NateBJones

9 күн бұрын

ooh that's one I haven't heard before

You’re picking out the rare success stories that Wall Street undervalued (Amazon AWS and Prime) and saying we’re at the same place with AI. What about all the failure stories like the Metaverse or Big Data or cryptocurrency or NFTs?

@NateBJones

9 күн бұрын

Fair question, and probably worth a follow-up video

AI has an energy problem: they run on massive amounts of fossil fuels. Ultimately unsustainable.

@dianapennepacker6854

13 күн бұрын

These power instability strongman arguments for EV and AI. Are old. Luckily renewables price keeps dropping amongst energy storage. Expect Solar to drop even more with Pervoskites allowing us to print solar cells. Some companies are already building their own power sources or contracting it out for power companies to hopefully do the same. Secondly the cat is out of the bag. There is no stopping progress, and it has become a nationality security agenda to lead AI. AI will help us in research and development paying for itself. Especially since computing power has still not bottlenecked. It will. Not for now. So the requirements for AI on the hardware side will get significantly cheaper, and as AI itself specializes it won't exactly require huge computers as CPU/GPUs get better. Lastly both will become more efficent eventually while renewable technology improves. Hell maybe cloud computing will become what it was supposed to become, and the technologies with 5g plus will merge to allow computing power to be more centralized. Also the power industry will receive a boom... For ages power demand didn't really go up. In fact for a good 7 years there till recently power demand was dropping. Now that demand is rising power companies will get investment both from the private and government sector. Creating more production to satisfy the demand. Yay for capitalism! (With that said some corporations will do corperation things, and pay stock holders instead of investing. Or government regulations like in CA, or stupidity like in TX will drag their feet on doing their fucking jobs. California with too many regulations, and Texas specifically declining on being connected to the two east/west grid, and having such lax regulations to where their grid is hot garbage.

@bennycarter5249

13 күн бұрын

@@dianapennepacker6854 Correct! We've known that fossil fuels are not renewable since the 70s and we are closer than ever to running out. That's a lot of maybe's; But here in the real world there is a big physical issue with the continuation of AI tenchonogy as it is is being fueled currently.

@dianapennepacker6854

13 күн бұрын

@@bennycarter5249 I think it is over valued like hell. That is the biggest issue. We don't need an AI controlled toothbrush or pair of socks. Like it is the wild west at the moment. Going to be a lot of failed companies in the next five years got a ton of investment for using buzz words. Still curious on when Nvidia will peak.. With that said the physical side of things like power demand, or resources will be fine. Fine as not the same, but all the issue aren't insurmountable at any stretch of the imagination. The grid needed a real rework anyway. Now we have reasons. Yet at the same time a lot of new technologies are coming out where just a few years of research will do wonders. Energy storage for example has so many options right now, with much that will get cheaper in the next two years, and I'm there are no clear winners that I'm aware of. I do however see losers. Gravity vault is a scam, and I feel like hydrogen is niche with too many issues like volumetric density being as terrible than my stupidity and intelligence - like you can make it work with a lot of effort in a pinch, but we can achieve greater things.

@postblitz

13 күн бұрын

@@dianapennepacker6854 "There is no stopping progress!" said person killing their environment. Wonderful thing "progress", it hits walls most marvelously and the exuberance is always at the peak just before the impact.

@nobodyspecial2053

13 күн бұрын

They are talking about energy requirements equivalent to major countries, needed to power this stuff and the development of the ever 20 years away fusion power to keep this stuff running. Not to mention, power isn't the only environmental issue, these thing are using up lakes of fresh water in or to cool (which can't be used again easily because chemical treatment).

unfortunately your audio is broken sounds like a record player needle that hops off the record every 3 seconds 🫠

@NateBJones

9 күн бұрын

Weird will have a look at that

@jnevercast

7 күн бұрын

@@NateBJones Seems fixed in "AI is not a chatbot" - thanks for the change in audio setup!