Live data from Hacker News

Satya Nadella: "We need to find something useful for AI"

pcgamer.com

161–170 of 228 posts

Re: Satya Nadella: "We need to find something useful for AI"

#161

As for me, ChatGPT and Claude were able to diagnose and fix health issues that multiple doctors failed to fix. I trust LLMs more than random doctors and blogs. Cause they are able to search the web deeply, search for up to date info/research and synergize all that. You can have back and fourth for as long as you need. The issue is that using LLMs properly requires a certain skill that more people lack.

I've come to distrust a number of "experts" in my time. It's not that I don't believe in expertise. It's that so many people bullshit it. From the people who promise to fix my drywall to the people who promise to fix my van. They are all fucking incompetent. I know because I've fixed the issues they have pointed out better than they have at a lower cost. Multiple times. Even when pricing out my own time at consultant rates.

And I don't mean I've just rejected the lowest offer to DIY shit to oblivion. I've accepted bids and been continuously disappointed with the results. To the point I no longer trust "experts" in these spaces because any "expertise" they bring is pretty shallow at best. The exception I'd point out is the auto-shop I prefer. They are busy enough that if I want to schedule something it's going to be 3-6 months out. As a result I've replaced my own suspension and replaced my alternator and starter myself while waiting for appointments with the rare actual experts in any domain. Actual expertise is rare and most folks don't know how to recognize it, especially outside of domains they are familiar with. Unfortunately it's way more profitable to fake expertise in various domains and collect payments and run than to actually stand behind your business and work. Thus the world we're forced to live in today.

Re: Satya Nadella: "We need to find something useful for AI"

#162

LLMs and their capabilities are very impressive and definitely useful. The productivity gains often seem to be smaller than intuitively expected though. For example, using ChatGPT to get a response to a random question like "How do I do XYZ" is much more convenient than googling it, but the time savings are often not that relevant for your overall productivity. Before LLMs you were usually already able to find the in…

I think you're underestimating how many people don't know how to properly search on google (i.e. finding the proper keywords, selecting the reputable results, etc etc). Those are probably also the same people that will blindly believe anything a LLM says unfortunately.

Re: Satya Nadella: "We need to find something useful for AI"

#163
post #157
post #130

Earlier quoted context omitted.

> We already know many useful things to do; there are already 10,000 startups (9789 out of YC alone, 4423 of which are coding-related) doing various ostensibly useful things. And there a ton more use-cases discussed in the comments here and elsewhere. But because of the headline the discussion is missing the much more important point! There has to be gold in the West! Look at all of the prospectors moving there to ge…

Cryptocurrency startups were just ways to bring "novel" trading strategies into the world of unregulated finance. The only "innovation" there was freedom from regulation and accountability, which turned out just as you'd expect. (Also there are still a surprising number of crypto companies plugging away out there.) "Uber but for ..." apps were either just bad ideas, or ended up serving niche markets or reincarnating…

> A lot of AI startups are already making a lot of money and growing at a record pace. Some of the numbers out there are bonkers.

Do you have examples of this? I'm aware of raises, but not aware of any profitable ai companies yet

Re: Satya Nadella: "We need to find something useful for AI"

#164
post #41
post #24

Earlier quoted context omitted.

Don't forget porn. Though it's a use case people like Satya will want to avoid for reasons.

"Why d'you think the net was born? Porn, Porn, Porn"

Adult conversation mode is just the start.

Re: Satya Nadella: "We need to find something useful for AI"

#166

LLMs and their capabilities are very impressive and definitely useful. The productivity gains often seem to be smaller than intuitively expected though. For example, using ChatGPT to get a response to a random question like "How do I do XYZ" is much more convenient than googling it, but the time savings are often not that relevant for your overall productivity. Before LLMs you were usually already able to find the in…

> For example, using ChatGPT to get a response to a random question like "How do I do XYZ" is much more convenient than googling it More convenient than traditional search? Maybe. Quicker than traditional search? Maybe not. Asking random questions is exactly where you run into time-wasting hallucinations since the models don't seem to be very good at deciding when to use a search tool and when just to rely on their t…

> I asked Gemini (generally considered one of the smartest models

Everyone is entitled to their own opinion, but I asked ChatGPT and Claude your XFCE question, and they both gave better answers than Gemini did (imo). Why would you blindly believe what someone else tells you over what you observe with your own eyes?

Re: Satya Nadella: "We need to find something useful for AI"

#167

Earlier quoted context omitted.

> For example, using ChatGPT to get a response to a random question like "How do I do XYZ" is much more convenient than googling it, but the time savings are often not that relevant for your overall productivity. Before LLMs you were usually already able to find the information quickly and even a 10x speed up does not really have too much of an impact on your overall productivity, because the time it took was already…

Why not have the LLM generate the library?

Because every time you run the LLM it will generate a new library, with new and surprising bugs.

It's better to take an existing, already curated and tested library. Which, yes, may have been generated by an LLM, but has been curated beyond the skill of the LLM.

Re: Satya Nadella: "We need to find something useful for AI"

#168
post #157

Earlier quoted context omitted.

Cryptocurrency startups were just ways to bring "novel" trading strategies into the world of unregulated finance. The only "innovation" there was freedom from regulation and accountability, which turned out just as you'd expect. (Also there are still a surprising number of crypto companies plugging away out there.) "Uber but for ..." apps were either just bad ideas, or ended up serving niche markets or reincarnating…

> A lot of AI startups are already making a lot of money and growing at a record pace. Some of the numbers out there are bonkers. Do you have examples of this? I'm aware of raises, but not aware of any profitable ai companies yet

I'm not sure about "profitable" because these are very young (2 - 3 years old) private companies, but there have been a few reports showing their growth indicating very strong PMF i.e. utility:

https://a16z.com/revenue-benchmarks-ai-apps/

https://www.cnbc.com/2025/03/15/y-combinator-startups-are-fa...

https://medium.com/@gjarrosson/ycs-revenue-explosion-497ea17...

https://stripe.com/blog/inside-the-growth-of-the-top-ai-comp...

https://www.ft.com/content/a9a192e3-bfbc-461e-a4f3-112e63d0b...

https://menlovc.com/perspective/2025-the-state-of-generative...

The revenue growth -- assuming these investors are not all colluding to fudge these numbers on a grand scale -- is way higher than what most have seen before.

Not everyone is hitting PMF of course, but apparently the success rate is also way higher than the past. Ignore the valuations and funding numbers, they are definitely inflated due to the hype.

Re: Satya Nadella: "We need to find something useful for AI"

#170
post #159

Earlier quoted context omitted.

Yes, but I think that is the exact attitude Satya's advising against: Where many leaders are looking inward to see how they can optimize their own businesses, they should be looking outward to see how they can improve the world.

LOL! Does anyone think CEO's of major corporations are trying to improve the world? Honestly that's funny. Sorry if you really are this earnest. I wish you the best and hope you can maintain your outlook despite how disconnected from reality.

Eh, I did say he's not necessarily better than the rest of them, just more aligned with the rest of us, likely because he's more forward-looking than the others. He's seeing the backlash against AI, the pushback against data centers, and the social turmoil on the horizon, and he's saying, "Maybe we should not piss all the people off."

Also: https://news.ycombinator.com/item?id=46729271

Post reply on HN