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

The Generative AI Con

wheresyoured.at

Post reply on HN
Print view View original

431–440 of 503 posts

Re: The Generative AI Con

#431

Post by daveguy »

scotty79 wrote: post #253
jcgrillo wrote: post #237

Earlier quoted context omitted.

> they’ll keep being used How? I get that many devs like using them for writing code. Personally I don't, but maybe someday someone will invent a UX for this that I don't despise, and I could be convinced. So what? That's a tiny market. Where in the landscape of b2b and b2c software do LLMs actually find market fit? Do you have even one example? All the ideas I've heard so far are either science fiction (just wait an…

> What is this shit for? Various minor thing so far. For example I heard about ChatGPT being evaluated as a tool for providing answers for patients in therapy. ChatGPT answers were evaluated as more empathetic, more human and more aligned with guidelines of therapy than answers given by human therapists. Providing companionship to lonely people is another potential market. It's not as good as people at solving proble…

Subjective evaluation of "humanity" and "empathy" in responses is much less important than clinical outcome. I don't think an online chat with a nebulous entity will ever be as beneficial as interactions that can, at least occasionally, be in-person. Especially as the trust of online conversations degrade. Erosion of trust online seems like a major negative consequence of all the generative AI slop (LLM or otherwise).

Re: The Generative AI Con

#432

Post by daveguy »

lukevp wrote: post #390
agentultra wrote: post #231

Earlier quoted context omitted.

That experience is heavily subsidized and is unprofitable for these companies providing it based on what we know. Even with all of the other developers who are also using the same work flow and espousing how great it is. Even with all of the monthly subscribers at various tiers. It has been unprofitable for several years and continues to be unprofitable and will likely remain unprofitable given current trends. The au…

Inference isn’t that expensive. A single junior dev costs orders of magnitude more than the amount of inference I use. Companies in growth mode don’t have to make money, it’s a land grab right now. But the expense is largely in the R and D. You can build a rig to run full models for 10-20k right? That’s only a month or two of a junior dev’s time, and after that it’s just electricity. And you could have dozens of devs…

Yeah, you can build a rig to run full models for 10-20k... That's a big reason OpenAI might not make it. The whole article is about LLMs not being a viable business.

Re: The Generative AI Con

#433

Post by crakhamster01 »

paulgb wrote: post #405
player1234 wrote: post #249

Earlier quoted context omitted.

I fail to see how a technology that is to expensive to maintain can have a killer app.

It’s only going to get cheaper over time. It’s already cheap enough that if these services disappeared overnight I’d switch to an open source alternative with a local model. The industry needed VC backing to pay the fixed cost of the research, but the cost of running inference is not insane compared to the volume it provides.

I think the argument this article makes though is that generative AI isn't generating enough value to justify the sky high valuations and investment. Sure, if all of these services disappeared overnight, the remaining users could self host or run a model locally. I feel like that speaks to the lack of value any one company provides though?

Maybe this era of AI will be remembered as a wealth transfer from VCs back to everyday consumers lol

Re: The Generative AI Con

#434

Post by Timwi »

jemmyw wrote: post #392
tdeck wrote: post #386

Earlier quoted context omitted.

To play devil's advocate here, it takes time for society to orient around new technologies and make them feel indispensable. If smartphones had disappeared in 2008, most users would be mildly inconvenienced. They'd go back to using a flip phone and a TomTom, or printing out map directions, and sending emails on their laptop with WiFi. No employer expected them to have a chat app on their phone* or use PagerDuty, no b…

> Perhaps in 5-10 years people will stop putting any effort into documentation or organizing information The LLM needs the documentation more than I do

LLMs (or other AIs) of the future may well get better than humans at reading badly written documentation, filling in gaps by cross-referencing multiple sources, or learning APIs entirely from reading uncommented source code...

Re: The Generative AI Con

#435

Post by synicalx »

lukevp wrote: post #12

I’m a little shocked at how much negativity there is around LLMs among developers. It’s a new tool that requires some learning, and it’s sometimes not so great, but if you’ve used an IDE with real coding assistance built in (eg. VS Code in Edit with Copilot mode - NOT Chat mode, using Claude 3.5), it’s honestly not much worse than a junior dev and 100x faster. And if the code is bad you throw it away and try again 10…

I don't really get the hate; work is boring, if some tool can make it happen faster and with less effort I'm all for it. I don't hear the line cooks at McDonalds complaining that they have to use a semi-automated grill that beeps at them instead of an open fire.

Re: The Generative AI Con

#436

Post by Timwi »

mitthrowaway2 wrote: post #288
pavlov wrote: post #105

Earlier quoted context omitted.

If Google search disappeared, I would still remember the names of several sites like Stack Overflow that are always at the top of the results, and I’d just go directly there. Maybe the original Yahoo! style curated list of categorized links would actually be more useful for me at this point than Google with all the SEO spam. That kind of high-quality directory combined with ChatGPT would probably replace Google for m…

Does anyone maintain a Yahoo-like directory anymore? I used to love browsing the internet that way, and I feel like it would be a really interesting way to find websites rather than content.

I agree, but I think such a project is just not realistic anymore. If you make it publicly editable (like a wiki) it'll just be full of spam, and if you don't, then you need swathes of human editors. And then you get into endless fruitless debates on exactly which websites should be listed and which should be excluded.

Re: The Generative AI Con

#437

Post by throw234234234 »

Aunche wrote: post #220

Protein folding is a an application of generative AI that will probably produce trillions of dollars of value in the long term. It was probably impossible for Google to squeeze that sort of money from researchers who use it, but it proves that the technology definitely useful. Another application that is highly underrated is with robots completing complicated tasks.

This is a case for example I wish AI was targeting. However its more likely they will build more benchmarks and target dev's/SWE's quickly and hard because that's probably what their VC's want them to do. The domains that will generally benefit society - that's more of a maybe they might do later.

They just released a benchmark today to try to attempt it (OpenAI).

Re: The Generative AI Con

#438

Post by Timwi »

lysecret wrote: post #161
levkk wrote: post #91

If my Android (or IPhone) disappeared tomorrow, I would feel like I time traveled back a century. If Google search was gone, I wouldn't be able to do my job anymore. If the cloud disappeared, I wouldn't be able to build apps anymore. There are no workarounds, unless you feel like going to a library...? If ChatGPT disappeared tomorrow (or derivatives like Copilot, etc.), I would be mildly inconvenienced. Then I'd go b…

Google search example is interesting I wouldn’t care at all. YouTube or maps on the other hand…

I've actually been wondering what platform(s) people would flock to if YouTube were to suddenly disappear. Vimeo? DailyMotion? Or even... PeerTube??...

Re: The Generative AI Con

#439

Post by steego »

bko wrote: post #391
light_triad wrote: post #336

If you've been in SV long enough, you've seen multiple hype cycles. AI is the latest one. It doesn't mean AI is a con, there's just a tendency to exaggerate and make hyperbolic claims to sell. AI is primarily used to write text and code. Actually integrating it into workflows across markets will take time. A great article that is neither overly pessimistic or optimistic is Benedict Evan's The AI Summer [1]. He argues…

The author betrays himself early with this line: > a cynical bubble inflated by OpenAI CEO Sam Altman built to sell into an economy run by people that have no concept of labor other than their desperation to exploit or replace it. He brings up the concept of labor and applies a moral judgement about "replacing" and "exploiting" labor. And then he throws the kitchen sink at the technology. People use it sure, but it's…

> So what is his problem?

My guess: He's just posting a hot-take to farm for engagement.

I'm not saying he's wrong about everything. I'm just pointing out that he has a small incentive to be engaging.

Re: The Generative AI Con

#440

Post by ROM_Bombadil »

pton_xd wrote: post #19
lukevp wrote: post #12

I’m a little shocked at how much negativity there is around LLMs among developers. It’s a new tool that requires some learning, and it’s sometimes not so great, but if you’ve used an IDE with real coding assistance built in (eg. VS Code in Edit with Copilot mode - NOT Chat mode, using Claude 3.5), it’s honestly not much worse than a junior dev and 100x faster. And if the code is bad you throw it away and try again 10…

I think the usefulness is just very domain specific. If you're writing some types of boilerplate or often-tutorialized code it can spit out something very reasonable. Other types of code, like say in game dev, it stumbles around and never produces anything usable. But like you said, in a few more years we'll see! It does feel like there's some missing pieces yet to be figured out to truly "reason" and generalize.

> If you're writing some types of boilerplate or often-tutorialized code it can spit out something very reasonable. Other types of code, like say in game dev, it stumbles around and never produces anything usable.

This makes me think of a quirk I discovered recently which is that ChatGPT simply won't generate a picture of a 'full glass of wine'. It generates pictures with all sorts of crazy waves/splashes in the glass but the glass is always half full no matter how you prompt it.

I'm not enough of an expert to make any deductions from this, but I think it hints at what the limitations of the currently models are.

Post reply on HN