In articles like this I’m always surprised that author did not take 5 minutes to think that well, maybe I’m living in a bubble and there’s lots of people that are actually excited about AI.
It must be truly abysmal everywhere else then, because it doesn't show much value on highly technical tasks when I try.
I think people who are interested in how things work and AI users are two entirely different cohorts. ChatGPT from user's perspective is more like a search engine or an autotranslator rather than some sophisticated technical gizmo.
Not sure if they are 'excited', but they are definitely using it.
Lots of interns and students also use the bots.
What I am annoyed by is having to tell users and management "no, LLMs can't do that" over and over and over and over and over. There's so much overhype and just flat out lying about capabilities and people buy into it and want to give decision making power to the statistics model that's only right by accident. Which: No.
It's a fun toy to play with and it has some limited uses, but fundamentally it's basically another blockchain: a solution in search of a problem. The set of real world problems where you want a lot of human-like writing but don't need it to be accurate is basically just "autocomplete" and "spam".
Today, 10 years ago crypto was what everyone wanted, you can see how bitcoin soared and crypto scams was everywhere and made many billions.
And no AI is not struggling to keep up with user demand, it is struggling to keep up with free but not paid demand. So what you mean is AI is struggling to keep up with investor demand, more people want to invest into AI than there are compute to buy, but that was the same for bitcoins, bitcoin mining massively raised the prices on GPU due to how much investors put into it.
But investor driven demand can disappear really quickly, that is what people mean with an AI bubble.
That's not even considering the probability that demand could slow as people lose interest.
Regarding people losing interest: are you willing to bet with me that there will be fewer than, let's say 500 million active users of LLMs in 5 years?
I don't think it's this. At least, I don't see a lot of that. What I do see a lot of is people realizing that AI is massively overhyped, and a lot of companies are capitalizing on that.
Until/unless it moves on from the hype cycle, it's hard to take it that seriously.
The average consumer does not appear to be particularly excited about products w/ AI features though. A big example that comes to mind is Apple Intelligence. It's not like the second coming of the iPhone, which it should be, given the insane amount of investment capital and press in the tech sphere.
Unless both the App Store and Google Play Store rankings are somehow determined primarily by HN users, then it seems like AI isn't only a thing on HN.
[0]: https://app.sensortower.com/overview/6448311069?tab=category...
And these people dont know a thing about C, Java, CPU or RAM. They are not tech people.
Over the decades the moment I hear real life conversation from non tech people in public talking about certain tech and being somewhat enthusiastic about it, is the moment that piece of tech has reached escape velocity. And it will go mainstream. And somewhat strangely enough, I only started using more ChartGPT because every non-tech people are starting to use it. And they use it much more than I do.
Just like people laughed about "Smartphone" as in iPhone era. Lots of tech people including i believe MKBHD only got their first Smartphone with iPhone 4, most consumer are even later. While I have watched the introduction of iPhone Keynote a dozen times before the thing was even shipped. The adoption curve of any tech will never be linear.
The example pointed out at the start of the article is somewhat bizarre. AWS is only pausing Colo. Apple Intelligence has more to blame with Apple themselves rather than AI. Intel ( or PC ) not selling AI enhanced chip is because consumer dont buy AI hardware, they buy AI functions. And so far nothing on Windows OS seems to be AI enhanced and specifically requires the AI Intel CPU.
And I am not even Pro AI or AI Optimist to see all that.
And they're right. They cited consumer research to show the ambivalence of consumers towards these products as well.
The general public do not appear to be great fans; it is arguably the _industry_ which is the bubble.
Moreover, slaves aren't productive over the medium/long term. You don't get useful and lengthy performance from raw coercion, the same way you don't get truthful and actionable intelligence from torturing prisoners.
So no, an LLM has very little in common with a slave.
No one is going to get a billion dollar investment due to that. It's why all the corporate speak and marketing is harping on about productivity, robot takeover and deus ex machina in corporate language.
Normal people will use it for creative writing aid, scrapbooking and other extremely unprofitable non-technical stuff.
Personally - i was able to get mockups of interior designs based on the photo of a building under construction using chatgpt - this would've cost me both time and money if i went to a real designer.
Gemini has been summarising meetings i could not attend (scheduling conflicts etc...) and saving me from hours of watching meeting recordings.
I was really skeptical of things because of the horrible results I've had 2 years ago with copilot and chatgpt, but things have improved drastically. To the point that it's already empowering certain people/jobs while having the opposite effect on others.
Is it perfect? Nope. The mockups did have weird glitches. But they were 75% there and good enough for the task I wanted. The meeting notes were as good as a real human.
So it's definitely eroding more of these kinds of jobs and so we are
That isn't pure doomerism - there's plenty of room for AI assist, and people like using AI experiences themselves. AI as a product is here to stay, but the second order of products openly using AI is showing it's limits.