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

31–40 of 503 posts

Re: The Generative AI Con

#31

Post by cmdtab »

ramraj07 wrote: post #22
cmdtab wrote: post #17

Earlier quoted context omitted.

Are they? I find Agentic mode on most editors barely useful. Autocomplete and inline editing is great though. To use these tools properly, you need to know how to build the same thing precisely.

But that still doesn't exclude copilot, which is the most intuitive and genuinely useful version of ai after chatgpt itself.

I don’t deny it.

I used claude + gemini to whip out a rewrite of a show HN project in less than half an hour to deployment yesterday.

https://news.ycombinator.com/item?id=43071381

But this sort of work is fairly low value and boilerplate-y.

Re: The Generative AI Con

#32

Post by epistasis »

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.

I keep on trying them, but if they are useful they are useful for only a small fraction of engineers at the moment. I'm not sure if this is due to the nature of the work, or the nature of the user.

I have heard "top" engineers at various places say it makes them 2x faster, or whatever, but I would like to see this assessed by timed testing, as is sometimes done for evaluating software engineering.

Copilot may let me type less, but I have not seen the wall clock effects, which is a very hard thing to measure (time perception is very unreliable).

Re: The Generative AI Con

#33

Post by CamperBob2 »

bccdee wrote: post #13
lukevp wrote: post #7

Why in the world would you want a car?? They are horrible to maintain, difficult to operate, expensive, smell bad, and are slower than horses! -somebody salty about automobiles, circa 1900.

Tbh they'd have been right. Cars are expensive to buy and to maintain. They are difficult to operate—people die every day. They do smell bad, and the particulate matter they emit is extremely bad for you. And they are slow: Traffic is an unsolvable problem in urban centres. Redesigning our cities around cars was one of the big mistakes of the 20th century.

It's our loss that you weren't around to set everyone straight, I'm sure.

Re: The Generative AI Con

#34

Post by advael »

xnx wrote: post #27

Plenty of AI companies that are cons or extremely overvalued, but the technology is the real deal and delivering huge improvements over previous techniques in all kinds of domains: language translation, weather prediction, code completion, self driving, etc.

I swear no matter how many times people say it people will still conflate all ML with LLMs. No, chatGPT is not driving advances in self-driving or weather prediction

Re: The Generative AI Con

#35

Post by ksynwa »

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…

He hasn't said anything about when the bubble will burst. Just says that it is a bubble and it will inevitably burst.

Re: The Generative AI Con

#36

Post by stuckkeys »

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…

Honestyl. It has its drawbacks but I am usually at 50x with few different agents running side by side. What we need is better GPU competition with tons of ram.

Re: The Generative AI Con

#37

Post by thfuran »

jayd16 wrote: post #18
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…

Unreliable tools are utterly exhausting. > not much worse than a junior dev and 100x faster. Is there a greater hell than this!?

Worse than an intern and 1000x faster?

Re: The Generative AI Con

#38

Post by cmdtab »

epistasis wrote: post #32
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.

I keep on trying them, but if they are useful they are useful for only a small fraction of engineers at the moment. I'm not sure if this is due to the nature of the work, or the nature of the user. I have heard "top" engineers at various places say it makes them 2x faster, or whatever, but I would like to see this assessed by timed testing, as is sometimes done for evaluating software engineering. Copilot may let me…

https://news.ycombinator.com/item?id=43071381

You can see this example where I timed myself to deployment using AI tools to rewrite a show HN project in half an hour. The code is open source.

My comment was posted 2 hours after show HN when I saw it on front page so you know I didn’t lose track of time I spent.

Re: The Generative AI Con

#39

Post by guyfromfargo »

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…

We’re already here. Check out Cline or Roocode. The technology is incredible.

I wrote a custom MCP to grab tickets from my Kanban board, Roo will pull down the tickets and start implementing them. I then have another agent that QAs, and either kicks the ticket back, or moves the ticket to human review.

I’m doing this on a real world micro SaaS. It has about 50 paying customers, and I’m the sole developer. I did a complete rewrite and the AI was able to complete about 90% of the project. I estimate I can get about a week’s worth of coding done in a day with this setup. I haven’t even scratched the surface of optimizing this workflow.

I’m also just one guy working nights and weekends, I’m sure there are many startups solving this same problem. It’s amazing to be shipping features this quick, but as a developer I’m terrified of what this is going to do to our careers.

Re: The Generative AI Con

#40

Post by idonotknowwhy »

simonw wrote: post #5

> When you put aside the hype and anecdotes, generative AI has languished in the same place, even in my kindest estimations, for several months, though it's really been years. The one "big thing" that they've been able to do is to use "reasoning" to make the Large Language Models "think" [...] This is missing the most interesting changes in generative AI space over the last 18 months: - Multi-modal: LLMs can consume…

> This is missing the most interesting changes in generative AI space over the last 18 months

I agree, though personally I'm liking the "big thing" as well. R1 is able to one-shot a lot of work for me, churning away in the background while I do other things.

> Multi-modal

IMO this is still early days and less reliable. What are some of your daily use cases?

> Context lengths

This is the biggest thing IMO (Models remaining coherent at > 32k contexts)

And whatever improvements have caused models like Qwen2.5 to be able to write valid code reliably vs the GPT-4 and earlier days.

There are a whole lot of useful smaller niche projects HF like extracting vocals/drums/piano from music, etc

Post reply on HN