Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A new AI winter is coming?

taranis.ie

Post reply on HN
Print view View original

161–170 of 301 posts

Re: A new AI winter is coming?

#161

Post by m4rtink »

Nevermark wrote: post #76

I am simply stunned at the negativity. Yes, there is hype. But if you actually filter it out, instead of (over) reacting to it in either direction, progress has been phenomenal and the fact there is visible progress in many areas, including LLMs, in the order of months demonstrates no walls. Visible progress doesn’t mean astounding progress. But any tech that is improving year to year is moving at a good speed. Huge…

Yeah, totally agreed - there is still far too little negativity in comparison to what is going on.

Re: A new AI winter is coming?

#162

Post by lm28469 »

tarr11 wrote: post #7

This has convinced many non-programmers that they can program, but the results are consistently disastrous, because it still requires genuine expertise to spot the hallucinations. I've been programming for 30+ years and now a people manager. Claude Code has enabled me to code again and I'm several times more productive than I ever was as an IC in the 2000s and 2010s. I suspect this person hasn't really tried the most…

Last week I gave antigravity a try, with the latest models and all, it generated subpar code that did the job very quickly for sure, but no one would have ever accepted this code in a PR, it took me 10x more time to clean it up than to have gemini shit it out.

The only thing I learned is that 90% of devs are code monkeys with very low expectations which basically amount to "it compiles and seems to work then it's good enough for me"

Re: A new AI winter is coming?

#163

Post by phantasmish »

injidup wrote: post #128

Most “AI is rubbish” takes treat it as an open-loop system: prompt → code → judgment. That’s not how development works. Even humans can’t read a spec, dump code, and ship it. Real work is closed-loop: test, compare to spec, refine, repeat. AI shines in that iterative feedback cycle, which is where these critiques miss the point.

What’s tough for me is figuring out where people are realizing significant improvements from this.

If you have to set up good tests [edit: and gather/generate good test data!] and get the spec hammered out in detail and well-described in writing, plus all the ancillary stuff like access to any systems you need, sign-offs from stakeholders… dude that’s more than 90% of the work, I’d say. I mean fuck, lots of places just skip half that and figure it out in the code as they go.

How’s this meaningfully speeding things up?

Re: A new AI winter is coming?

#164

Post by WhyOhWhyQ »

dangus wrote: post #57
john61 wrote: post #49

Earlier quoted context omitted.

> The argument that computational complexity has something to do with this could have merit but the article certainly doesn’t give indication as to why. OP says it is because that predicting the next token can be correct or not, but it always looks plausible because that is what it calculates. Therefore it is dangerous and can not be fixed because it is how it works in essence.

I just want to point out a random anecdote. Literally yesterday ChatGPT hallucinated an entire feature of a mod for a video game I am playing including making up a fake console command. It just straight up doesn’t exist, it just seemed like a relatively plausible thing to exist. This is still happening. It never stopped happening. I don’t even see a real slowdown in how often it happens. It sometimes feels like the o…

Another anecdote. I've got a personal benchmark that I try out on these systems every time there's a new release. It is an academic math question which could be understood by an undergraduate, and which seems easy enough to solve if I were just to hammer it out over a few weeks. My prompt includes a big list of mistakes it is likely to fall into and which it should avoid. The models haven't ever made any useful progress on this question. They usually spin their wheels for a while and then output one of the errors I said to avoid.

My hit/miss rate with using these models for academic questions is low, but non-trivial. I've definitely learned new math because of using them, but it's really just an indulgence because they make stuff up so frequently.

Re: A new AI winter is coming?

#165

Post by palmotea »

anthonypasq wrote: post #130
cmiles8 wrote: post #84

Earlier quoted context omitted.

Correct

wrong. OpenAI is literally the only AI company with horrific financials. You think google is actually bleeding money on AI? they are funding it all with cash flow and still have monster margins.

> You think google is actually bleeding money on AI? they are funding it all with cash flow and still have monster margins.

They can still be "bleeding money on AI" if they're making enough in other areas to make up for the loss.

The question is: "Are LLMs profitable to train and host?" OpenAI, being a pure LLM company, will go bankrupt if the answer is no. The equivalent for Google is to cut its losses and discontinue the product. Maybe Gemini will have the same fate as Google+.

Re: A new AI winter is coming?

#166

Post by sidibe »

Nevermark wrote: post #76

I am simply stunned at the negativity. Yes, there is hype. But if you actually filter it out, instead of (over) reacting to it in either direction, progress has been phenomenal and the fact there is visible progress in many areas, including LLMs, in the order of months demonstrates no walls. Visible progress doesn’t mean astounding progress. But any tech that is improving year to year is moving at a good speed. Huge…

I have pretty negative feelings about all this stuff and how the future will be but also have to admit it's crazy how good it is at so many things I would have considered safe a few years ago before chatgpt.

There are a couple really disingenuous bloggers out there who have big audiences themselves and are "experts" for others audiences who really push hard this narrative that AI is a joke and will never progress by where it is today, it is actually completely useless and just a scam. This is comforting for those of us that worry more than are excited about AI so some eat it up while barely trying it for themselves

Re: A new AI winter is coming?

#167

Post by lamontcg »

stanfordkid wrote: post #16

This article uses the computational complexity hammer way too hard, discounts huge progress in every field of AI outside of the hot trend of transformers and LLMs. Nobody is saying the future of AI is autoregressive and this article pretty much ignores any of the research that has been posted here around diffusion based text generation or how it can be combined with autoregressive methods… discounts multi-modal model…

> that enables a computer to pretty effectively understand natural language

I'd argue that it pretty effectively mimics natural language. I don't think it really understands anything, it is just the best madlibs generator that the world has ever seen.

For many tasks, this is accurate 99+% of the time, and the failure cases may not matter. Most humans don't perform any better, and arguably regurgitate words without understanding as well.

But if the failure cases matter, then there is no actual understanding and the language the model is generating isn't ever getting "marked to market/reality" because there's no mental world model to check against. That isn't going to be usable if there are real-world consequences of the LLM getting things wrong, and they can wind up making very basic mistakes that humans wouldn't make--because we can innately understand how the world works and aren't always just stringing words together that sound good.

Re: A new AI winter is coming?

#168

Post by fragmede »

stanfordkid wrote: post #16

This article uses the computational complexity hammer way too hard, discounts huge progress in every field of AI outside of the hot trend of transformers and LLMs. Nobody is saying the future of AI is autoregressive and this article pretty much ignores any of the research that has been posted here around diffusion based text generation or how it can be combined with autoregressive methods… discounts multi-modal model…

> I could see many arguments about why modern research will fail to create AGI

Why is AGI even necessary? If the loop between teaching the AI something, and it being able to repeat similar enough tasks; if that loop becomes short enough, days or hours instead of months, who cares if some ill-defined bar of AGI is met?

Re: A new AI winter is coming?

#169

Post by matthewkayin »

frotaur wrote: post #35
gishh wrote: post #21

Earlier quoted context omitted.

> something fundamental has changed that enables a computer to pretty effectively understand natural language. You understand how the tech works right? It's statistics and tokens. The computer understands nothing. Creating "understanding" would be a breakthrough. Edit: I wasn't trying to be a jerk. I sincerely wasn't. I don't "understand" how LLMs "understand" anything. I'd be super pumped to learn that bit. I don't…

It astonishes me how people can make categorical judgements on things as hard to define as 'understanding'. I would say that, except for the observable and testable performance, what else can you say about understanding? It is a fact that LLMs are getting better at many tasks. From their performance, they seem to have an understanding of say python. The mechanistic way this understanding arises is different than huma…

I think it is fair to say that AIs do not yet "understand" what they say or what we ask them.

When I ask it to use a specific MCP to complete a certain task, and it proceeds to not use that MCP, this indicates a clear lack of understanding.

You might say that the fault was mine, that I didn't setup or initialize the MCP tool properly, but wouldn't an understanding AI recognize that it didn't have access to the MCP and tell me that it cannot satisfy my request, rather than blindly carrying on without it?

LLMs consistently prove that they lack the ability to evaluate statements for truth. They lack, as well, an awareness of their unknowing, because they are not trying to understand; their job is to generate (to hallucinate).

It astonishes me that people can be so blind to this weakness of the tool. And when we raise concerns, people always say

"How can you define what 'thinking' is?" "How can you define 'understanding'?"

These philosophical questions are missing the point. When we say it doesn't "understand", we mean that it doesn't do what we ask. It isn't reliable. It isn't as useful to us as perhaps it has been to you.

Re: A new AI winter is coming?

#170

Post by Quinner »

marknutter wrote: post #153
beachtaxidriver wrote: post #93

Earlier quoted context omitted.

It's true, but do you really trust the AI generated + Nurse Review output more than Organic Nurse generated? In my experience, management types use the fact that AI generated + Nurse Review is faster to push a higher quota of forms generated per hour. Eventually, from fatigue or boredom, the human in the loop just ends up being a rubber stamper. Would you trust this with your own or your children's life? The human in…

These same questions could be asked about self driving cars, but they've been shown to be consistently safer drivers than humans. If this guy is getting consistently better results from ai+human than it is from just humans, what would it matter if the former results in errors given the latter results in more and costs more?

If the cars weren't considerably safer drivers than humans they wouldn't be allowed on the road. There isn't as much regulation blocking deploying this healthcare solution... until those errors actually start costing hospitals money from malpractice lawsuits (or not), we don't know whether it will be allowed to remain in use.
Post reply on HN