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

211–220 of 503 posts

Re: The Generative AI Con

#211

Post by sandos »

cookiemonsieur wrote: post #117
dzonga wrote: post #69

for people using these AI products for coding what exactly are you doing ? general API plumbing ? call this api, combine the json results & spit it out. Then slap a React / Next.js frontend ? lately, I have been doing your classical business apps - due to the domain rules - ai is pretty useless - but I have found perplexity and deep mind to be smarter stack overflows. that's it.

personally I use it as a glorified stack overflow. It really just saves me a few clicks on my browser. For anything even remotely complex or involved, it's just not that useful. Interestingly, It seems to shine the most when doing boilerplate stuff in widely used languages such as python or javascript, but it's extremely bad with terraform.

Its very smart templates imo.

Its basically more time-consuming for me to get it to where I want the code than finish it myself. For some scaffolding in python green-field development, sure I save a few minutes.

Re: The Generative AI Con

#212

Post by adamc »

Aeolun wrote: post #206
danroblew wrote: post #81

Earlier quoted context omitted.

The article is about how the economics of the LLM market is making all tech look bad. They need trillions of dollars in returns. VC's won't finance tech startups for decades. I use Cursor sometimes, and VSCode + Continue with llama.cpp, and it's great. That's not worth billions. It's definitely not worth trillions.

> That's not worth billions. It's definitely not worth trillions. That is a problem for the VC’s that bet wrong, not for the world at large. The models exist now and they’ll keep being used, regardless of whether a bunch of rich guys lost a bunch of money.

Their ongoing operation is quite expensive, so even that is not assured.

Re: The Generative AI Con

#213

Post by trashtester »

surgical_fire wrote: post #118
roenxi wrote: post #107

Earlier quoted context omitted.

> I’m a little shocked at how much negativity there is around LLMs among developers. While the timeline is unclear; it seems likely that LLMs will obsolete precisely the skills that developers use to earn their income. I imagine a lot of them feel rather threatened by the rapid rate of progress. Pointing out that it is already operating at junior dev quality and rapidly improving is unlikely to quiet the discontent.

I use LLMs in coding. There are Junior Devs in my team. If you think LLMs operate at "junior dev" capacity you either don't work with junior devs and is just bullshitting your way around here, or you just pick pretty awful junior devs. LLMs are alright. An okay productivity tool, although its inconsistencies many times nullify productivity gains - By design they often spit out wrong results that look and sound very p…

If you define "junior" based mostly on age, then LLM's aren't yet at the level of a good "junior".

If you base it on ability, then an LLM can be be more useful to a good developer than 1 or more less competent "junior" team members (regardless of their age).

Not because it can do all the things like any "junior" can (like make coffee), but because the things it can do on top of what a "junior" can do, more than makes up for it.

Re: The Generative AI Con

#214

Post by paulgb »

drpossum wrote: post #158
paulgb wrote: post #11

> And even then, we still don't have a killer app! There is no product that everybody loves, and there is no iPhone moment! I would strongly argue that coding assistants are AI’s first killer app. Copilot, Cursor, Windsurf etc.

Say you reduced their revenue to only that application. Would it be sustainable? Would it be worth the billions upon billions of dollars that have been shoveled at it? Would it add more than the billions upon billions of dollars in the end? By your logic I could claim a quantum computer with qubits on the scale of the mass of the sun is a killer app for doing RSA encryption breaking. And I would be making an equally…

This is moving the goalpost on what "killer app" means. Code assistants are a compelling use of the tech that has quickly shown real-world value, which is the point I'm trying to make here.

Whether the companies that are leading the market today will end up being the ones who capture that value is anyone's bet.

Re: The Generative AI Con

#215

Post by camdenreslink »

Ancapistani wrote: post #41
locusofself wrote: post #4

As a 40yo software engineer (at Microsoft) with no specific domain expertise other than using genAI for fun and some code completion, this essay/blog post articulates my gut feelings about where we are at, very well.

41 here, working in healthtech… and Devin has committed more code and closed more tickets on my behalf in the past week at my behest than I’ve done on my own in a month. It’s basically functioning as a team of entry-level junior engineers at this point. Previously I was having to spend a fair amount of time writing tickets and providing context, but lately I’ve fed all my meeting transcripts and such into an LLM and…

I would be very curious about the size and complexity of this codebase. Every review of Devin I’ve seen has been very negative (burns a ton of money, gets stuck, doesn’t implement the changes you want).

For large codebases (greater than 15k or 20k LOC) the context size seems like a real problem right now.

Re: The Generative AI Con

#216

Post by twobitshifter »

bccdee wrote: post #10
anothermathbozo wrote: post #3

This is one of the most tilted pieces I’ve ever read. For months Ed has predicted The AI Bubble will burst “any day now” frequently citing ai company’s revenue as a sign the product is not viable and the valuations are too high. The valuations seem to be primarily based on the R&D progress instead of on a theory that widespread adoption of the existing product will experience an uptick. The current landscape imho sho…

> The valuations seem to be primarily based on the R&D progress There hasn't been much R&D progress, though. Sure, as another commenter pointed out, context lengths have gotten longer and chat models can interpret images now, but the industry figureheads have been pushing agents, and we're not much closer to those than we were two years ago when GPT-4 came out. Current models simply are not consistent enough to do th…

Recent results are showing exponential improvement in reasoning and dramatic decreases in the time and cost to train models. O3 now ranks 50th on code forces according to openai staff. Are you aware of all of this and still say R&D hasn’t progressed?

Re: The Generative AI Con

#217

Post by nprateem »

danroblew wrote: post #81
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…

The article is about how the economics of the LLM market is making all tech look bad. They need trillions of dollars in returns. VC's won't finance tech startups for decades. I use Cursor sometimes, and VSCode + Continue with llama.cpp, and it's great. That's not worth billions. It's definitely not worth trillions.

Yeah whatever. VCs will keep backing entrepreneurs, that's their job. Until there's a better way to get 10-100x returns, we're fine.

Re: The Generative AI Con

#218

Post by hoseja »

automatic6131 wrote: post #87
danroblew wrote: post #81

Earlier quoted context omitted.

The article is about how the economics of the LLM market is making all tech look bad. They need trillions of dollars in returns. VC's won't finance tech startups for decades. I use Cursor sometimes, and VSCode + Continue with llama.cpp, and it's great. That's not worth billions. It's definitely not worth trillions.

This is the crux. A cool thing has been invented, with real usages. Unfortunately, it's cost hundreds of billions of dollars and it has absolutely zero hope of making the trillions needed to justify that. Now someone will respond about how it's just a stepping stone, and how the billions are justified by _something completely imaginary, and not invented yet, and maybe not ever_ e.g. agents.

I see it a little differently. What was the direct economic return of the Manhattan Project?

Re: The Generative AI Con

#219

Post by Kappa90 »

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…

If you want to see an LLM tackling unit testing, check https://www.codebeaver.ai

Disclaimer: I’m the developer behind CodeBeaver

Re: The Generative AI Con

#220

Post by Aunche »

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.
Post reply on HN