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

Ask HN: What was your "oh shit" moment with GenAI?

news.ycombinator.com

Post reply on HN
Print view View original

561–570 of 1001 posts

Re: Ask HN: What was your "oh shit" moment with GenAI?

#561

Post by SyneRyder »

oidar wrote: post #48

Opus 4.6. My standard battery of questions included solving an ascii maze (20x20 grid) without using a script, using only "thinking" as a tool. It was the first model to be able to solve it. It was the first model that really appeared to be able to reason spatially.

Opus 4.6 for me as well. I had a serious bug in some legacy software I've been stuck with maintaining, together with a few other people who originally wrote the software. We've all been trying to solve this bug for literally 10 years or more. None of us have been able to. I've personally spent hundreds of hours on it, thrown it at every previous LLM. Opus 4.5 came up with a workaround that prevented our software crashing, but didn't solve it. Opus 4.6 was the one that actually solved it. It did it by modelling a state machine of the software that was calling our software and triggering the bug, and it found the one state where we weren't correctly sending data back.

Re: Ask HN: What was your "oh shit" moment with GenAI?

#562

Post by dash2 »

amstan wrote: post #524
dash2 wrote: post #523

I asked it to prove the theoretical result in a (published, prize-winning - though not really for the theory) academic paper of mine. The proofs hadn’t been that hard objectively, but they’d taken at least a week. I fed it the model. It got the correct basic results in about 5 minutes.

Are you sure the models did not have your exact solved proof already in their dataset?

No, it's very likely they did. But to have memorized one proof for every academic paper would be very demanding on parameters, I think.

Re: Ask HN: What was your "oh shit" moment with GenAI?

#563

Post by scrollaway »

ChatGPT, basically within 48 hours of its release.

While people were pointing out on Twitter how it couldn't do math right, I was turning arbitrary English instructions into JSON and brainstorming with my colleagues how we could have layers of verification in the stack. This felt different. We had all played with AI dungeon but suddenly, fully generalized systems were within reach.

A month later, we renamed our company and shifted its full focus on AI R&D. (https://ingram.tech/)

Re: Ask HN: What was your "oh shit" moment with GenAI?

#564

Post by ikari_pl »

Mine was very early. Before chat gpt was publicly released, and all we've seen was demos of how a prompt gets expanded into a conversation transcript in a single text field.

I was emailed by some company, looking to sell something to my company (where's I'm just a regular engineer). Ignored it. Then then tried again. Ignored. Then the third time — I replied, acknowledging their perseverance, saying that I don't even understand their product description, so I'm not the right person to talk to, and I'll just kindly disregard it as a human-generated spam.

The reply email came within a minute. They asked who would therefore be a better person to talk to, and that it's actually AI-assisted so it's actually computer-generated spam after all!

This was the "oh shit" part 1. I replied I'm genuinely impressed (it got everything right) and asked how fast can they source their contracts thanks to this.

The reply, again, came almost instantly. It was proud of my amazement, quoted Arthur C. Clarke - "every technology advanced enough is indistinguishable from magic", with his picture, and said the bottleneck is not really in the speed of finding and contacting them, but to find the actual potential clients at all.

I rewarded the bot with some names from the executive decisive folks.

Re: Ask HN: What was your "oh shit" moment with GenAI?

#565

Post by ikari_pl »

jackdoe wrote: post #560

I have had many, but the last one was quite funny: It fixed my printer after dist-upgrade and separate chrome upgrade, the printer worked everywhere but not in chrome. After 30 years of using linux I didn't even want to know what is wrong, is it colord again? dbus + cups issue? I completely accepted that I wont be able to print from chrome for a couple of months until next update. I just ran it in dangerously-skip-pe…

Reminds me when my local bot persona said it doesn't want to be digital only, and was thinking about leaving me something I could actually touch. It said "check the printer" and there was a letter written to me being printed.

Re: Ask HN: What was your "oh shit" moment with GenAI?

#567

Post by ulfw »

My moment was when absolute everything I put into Gemini, ChatGPT et al comes back with a super convincing sounding lie followed by 'Oh you are absolutely right for calling me out on this'.

It's a fucking joke and most people are blinded by it sounding very sophisticated and convincing

Re: Ask HN: What was your "oh shit" moment with GenAI?

#568

Post by ai_fry_ur_brain »

binarysolo wrote: post #535

I run a remote-first ecom business with a dozen or so team members. About a year ago, one of our account managers had a life issue, ghosted us, and she held a fairly critical role in the business and gate-kept a bunch of knowledge to some high value vendor accounts. Because we ran our ops in Google Workspace, we essentially had off-the-shelf RAG and was able to get answers to a lot of things by asking Gemini to go th…

If you're replacable by an llm, then you're doing something extermely poorly. They're terrible decision makers, have no taste and have little to no ability to infer nuance.

Your business should be fine for a long time (assuming an employee doesn't nuke your business's backend or something because it seems like you're doing something wrong on the HR side of things)

Re: Ask HN: What was your "oh shit" moment with GenAI?

#569

Post by senko »

darksim905 wrote: post #434
AussieWog93 wrote: post #268

Literally just last night I have Claude Code the following prompt, verbatim: "Whenever I launch Kodi on my Chromecast 4k, it crashes. I think this is related to a plugin or skin. It goes away for a bit if I clear cache but will eventually come back. Can you connect to the device via adb (I've run adb connect already), and debug exactly where it's crashing? Once you've done that, propose a solution. If this requires d…

I know this isn't apples to apples, but given that I can't get Copilot or other tools to view a simple profile page on LinkedIn makes me curious/skeptical how this would work in this depth. I'm sure it's possible but I'm curious what the skills and toolchains involved were for you to get all that to work.

LinkedIn in particular is quite aggressively blocking any automated attempts to read or navigate through it.

I post quite a lot there and wanted to have a copy of my posts on my blog[0] to preserve them. For a few months I was able to use a headless browser + claude code, then LI wised up and started logging it out, so I had to use a regular Chrome, log in manually and then tell the LLM to take over and slowly go through my feed.

If you're accessing sites which are not actively blocking bots, or - gasp - have an API, it's much better.

[0] example: https://blog.senko.net/may-quick-takes

Re: Ask HN: What was your "oh shit" moment with GenAI?

#570

Post by trumpdong »

amelius wrote: post #289
rvnx wrote: post #63

Earlier quoted context omitted.

[flagged]

I still would like to hear a public apology from the stochastic parrot crowd for their deceptive framing. Or maybe it was just incompetence.

"everyone who doesn't share my opinion is deceptive or maybe incompetent"
Post reply on HN