News for hackers. Stuff that matters.
Re:In 2025, venture capital can't pretend everything is fine... by calebkaiser on Sunday May 11, 2025 @11:50AM (#43954595)
I understand the spirit in this line of criticism, but I think it's easy to muddle the timelines and feel as if things "aren't moving," when in fact, the pace of research and improvement is great.

For context:

- GPT 2 was released in Feb 2019

- GPT 3 came out roughly 18 months later in 2020. It was a huge jump, but still not "usable" for many things.

- InstructGPT came out roughly 18 months later in early 2022, and was a huge advancement. This is RLHF's big moment.

- About 10 months later, ChatGPT is released at the end of 2022 as a "sibling" to InstructGPT. It's an "open research preview" at this point. This is around the time OpenAI starts referring to certain models as being in the "3.5 family"

- GPT-4 comes out in March 2023, so barely 2 years ago now. Huge jumps in performance, context window size, and it supports images. This is around the time ChatGPT hits 100 million users and is really becoming a reliable, widely adopted tool. This is also the same time that tools like Cursor are hitting the market, though they haven't exploded yet. Models are just now getting "good enough" for these kinds of applications

- GPT-4-Turbo comes out in November 2023, with way larger context windows and lower pricing.

- About 12 months ago, GPT-4o released, showing slightly increased performance on existing benchmarks over 4, but now with state-of-the-art audio capability support for something like 50 languages.

- 5 months ago, o1 releases. This is a big moment for scaling compute at test time, which is a major current research direction in ML. Shows huge improvements (something like 8x over 4) on some math/reasoning benchmarks. Within months, we have o3 and o4, which substantially improve these scores even further.

- In February of this year, we get 4.5, and then months later, the confusingly named 4.1, which shows improvements over 4o.

So to be clear, in 2019 we had an interesting research project that only a few people could tinker with.

18 months later, we had a better model that you could play with via an API, but was still a toy.

It takes more than two years to go from that to ChatGPT, and a few more months (nearly 3 years total) to get to the "useful" version of ChatGPT that really sets the world on fire. It took roughly 4 + 1/2 years to go from "novelty text generation" to "useful text generation".

In the 2 years since then, we've gotten multimodal models, a new class of reasoning models, baseline improvement across performance, and more. If anything, there is more fundamental research and wider variety of directions now (the kind of stuff that shifts paradigms) than before.

Re:In 2025, venture capital can't pretend everything is fine... by bigstrat2003 on Sunday May 11, 2025 @12:23PM (#43954823)
I don't agree. I think it's still just a toy that can't get anything useful done. It still hallucinates quite a bit, and gives confidently wrong answers all the time. It's impressive that we could make a program that does all those things, but until they get the accuracy to a reasonable level it won't be useful.
Re:In 2025, venture capital can't pretend everything is fine... by NitpickLawyer on Sunday May 11, 2025 @12:33PM (#43954907)
> I don't agree. I think it's still just a toy that can't get anything useful done.

Yeah, sorry, but your view is not supported by data. There are billions of $ spent on tokens every year (between oAI, anthropic, goog, etc). This is not a toy anymore. People are using it, one way or another so it's useful to them - to the tune of billions of dollars. How useful it'll turn out to be in the abstract is still up for debate, but it is useful today.

Re:In 2025, venture capital can't pretend everything is fine... by ghaff on Sunday May 11, 2025 @12:43PM (#43954990)
Companies are willing to spend money on it today which isn't the same thing. Personally, I probably wouldn't spend a penny on it (OK, maybe a dollar) but really haven't found generative AI to be particularly useful.
Re:In 2025, venture capital can't pretend everything is fine... by NitpickLawyer on Sunday May 11, 2025 @12:50PM (#43955035)
Companies usually go for "business gpt" in azure or whatever oAI & MS are offering. I'm talking about regular users at 20$ a pop. Unless all the providers are lying, billions of $ are being spent right now, by regular users. You may not find it useful, but others are.
Re:In 2025, venture capital can't pretend everything is fine... by ghaff on Sunday May 11, 2025 @12:58PM (#43955097)
Possibly unless it's just FOMO because Bill at the next desk is using it. I have tried some of the products. Just didn't find them more than casually useful. But I obviously can't speak for everyone's experiences. And different people have different needs.
Re:In 2025, venture capital can't pretend everything is fine... by ripe on Sunday May 11, 2025 @01:00PM (#43955105)
Some of that money is these companies trying to find a use for this technology, because their investors don't want them to miss the AI train. Not sure what proportion is seeing any returns right now.

If the product were to disappear tomorrow, I doubt the real economy would notice the loss of 0.05% in productivity.

[1] https://www.technologyreview.com/2025/02/25/1111207/a-nobel-...

Re:In 2025, venture capital can't pretend everything is fine... by bandrami on Sunday May 11, 2025 @01:05PM (#43955148)
Have those billions of dollars produced a single viable product?
Re:In 2025, venture capital can't pretend everything is fine... by NitpickLawyer on Sunday May 11, 2025 @02:11PM (#43955703)
Is this in jest? Chatgpt in itself is a viable product (based on how many active paying users they have). And then there's code w/ the myriad of implementations. Unless you've been hiding under a rock since 2023, we went from "cute toy, look it even looks like python" to "well, this agent-built, zero intervention, scribbled notes prompt to full-stack application probably has some security bugs, so be careful" real quick.
Re:In 2025, venture capital can't pretend everything is fine... by bandrami on Sunday May 11, 2025 @07:15PM (#43958099)
OK, so where are these applications it has built? What’s a good case study I can look at?
Re:In 2025, venture capital can't pretend everything is fine... by crowbahr on Wednesday May 14, 2025 @07:11AM (#43983100)
ChatGPT loses money on every single subscriber, even the $200/mo ones.

That's not a viable business - that's still a VC funded growth mode startup.

Re:In 2025, venture capital can't pretend everything is fine... by calebkaiser on Sunday May 11, 2025 @02:13PM (#43955717)
The number of software engineers using IDEs like Cursor is staggering.

Among knowledge workers in general, ChatGPT is used widely for basically any task that requires writing or researching.

This is obviously not going to be true for every single knowledge worker in every single role, and it seems that you don't find it particularly useful, but the volume of paying users is hard to dismiss out of hand.

Re:In 2025, venture capital can't pretend everything is fine... by bandrami on Sunday May 11, 2025 @06:44PM (#43957860)
I don't doubt it's widely used, what I don't see is any productivity increase among knowledge workers since its adoption. Are you aware of any signs of that?
Re:In 2025, venture capital can't pretend everything is fine... by threecheese on Sunday May 11, 2025 @01:14PM (#43955224)
I spend less time on web search, I suppose. But in trade I need to pay for it, while still providing my behavior data to the surveillance capitalism platforms.

It’s also improved my ability to analyze data, generating graphs and insights. But I still need to run that last mile myself, because I can’t fully trust its output. Same for web search actually, when I have a need to be comprehensive.

Re:In 2025, venture capital can't pretend everything is fine... by bandrami on Sunday May 11, 2025 @02:10PM (#43955690)
But has that resulted in you making a commercially viable product?

That's the step that still seems to be missing

Re:In 2025, venture capital can't pretend everything is fine... by threecheese on Monday May 12, 2025 @06:56PM (#43968236)
If AI companies can save $2 to pocket $.01 - across the entire market of human effort - then we have choice and no stake. We arent the customer, we’re the tangible good.
Re:In 2025, venture capital can't pretend everything is fine... by bandrami on Monday May 12, 2025 @10:48PM (#43969280)
Why would we need "AI companies"? Isn't this supposed to supercharge individual developers? But all I see people making with it is tooling for AI systems. Literally nobody's made a text editor or MIDI sequencer with it and just put the result on Github?
Re:In 2025, venture capital can't pretend everything is fine... by jcgrillo on Sunday May 11, 2025 @01:01PM (#43955118)
Companies spending money on SaaS means, generally, that a detached executive many rungs away from the actual users was sufficiently wowed by a demo/steak dinner/whatever to sign a deal. It has nothing in general to do with utility. To get an industry to spend billions you need only target a few hundred or a few thousand people.