> 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 negligible.

I'd even question that. The pre-LLM solutions were in most cases better. Searching a maintained database of curated and checked information is far better than LLM output (which is possibly bullshit).

Ditto to software engineering. In software, we have things call libraries: you write the code once, test it, then you trust it and can use it as many times as you want forever for free. Why use LLM generated code when you have a library? And if you're asking for anything complex, you're probably just getting a plagiarized and bastardized version of some library anyway.

The only thing where LLMs shine is a kind of simple, lazy "mash this up so I don't have to think about it" cases. And sometimes it might be better to just do it yourself and develop your own skills instead of use an LLM.

If only search engine AI output didn't constantly haluciate nonexistent APIs, it might be a net productivity gain for me...but it's not. I've been bit enough times by their false "example" output for it to be a significant net time loss vs using traditional search results.

The real benefit to a search engine is to rework and launder other people's information and make it your information.

Now instead of the wikipedia article you are reading the exact same thing from google's home page and you don't click on anything.

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.

LLM output is quickly rendering google search unusable, so it's kind of creating its own speedup multiplier.

It really depends on what 'XYZ' is and how many hoops you need to jump through to get to the answer. ChatGPT gets information from various places and gives you the answer as well as the explanation at each step. Without tools like ChatGPT its definitely not negligible in a lot of cases.

I use ChatGPT “thinking” mode as a way to run multiple searches and summarize the results. It takes some time, but I can do other stuff in another tab and come back.

It’s for queries that are unlikely to be satisfied in a single search. I don’t think it would be a negligible amount of time if you did it yourself.

This is partly because Google is past the enshittification hump and ChatGPT is just starting to climb up it - they just announced ads.

The difference is that in the past that information had to come from what people wrote and are writing about, and now it can come from a derivative of an archive of what people once wrote, upon a time. So if they just stop doing that — whether because they must, or because they no longer have any reason to, or because they are now drowned out in a massive ocean of slop, or simply because they themselves have turned into slopslaves — no new information will be generated, only derivative slop, milled from derivative slop.

I think we all understand that at this point, so I question deeply why anyone acts like they don’t.

That makes me think about the development of much software out there: the development time is often several orders of magnitude smaller than its life cycle.

> 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 training data.

For example, just now I was asking Gemini how to fix a bunch of Ubuntu/Xfce annoyances after a major upgrade, and it was a very mixed bag. One example: the default date and time display is in an unreadably small "date stacked over time" format (using a few pixel high font so this fits into the menu bar), and Gemini's advice was to enable the "Display date and time on single line" option ... but there is no such option (it just hallucinated it), and it also hallucinated a bunch of other suggestions until I finally figured out what you need to do is to configure it to display "Time only" rather than "Data and Time", then change the "Time" format to display both data and time! Just to experiment, I then told Gemini about this fix and amusingly the response was basically "Good to know - this'll be useful for anyone reading this later"!

More examples, from yesterday (these are not rare exceptions):

1) I asked Gemini (generally considered one of the smartest models - better than ChatGPT, and rapidly taking away market share from it - 20% shift in last month or so) to look at the GitHub codebase for an Anthropic optimization challenge, to summarize and discuss etc, and it appeared to have looked at the codebase until I got more into the weeds and was questioning it where it got certain details from (what file), and it became apparent it had some (search based?) knowledge of the problem, but seemingly hadn't actually looked at it (wasn't able to?).

2) I was asking Gemini about chemically fingerprinting (via impurities, isotopes) roman silver coins to the mines that produced the silver, and it confidently (as always) comes up with a bunch of academic references that it claimed made the connection, but none or references (which did at least exist) actually contained what it claimed (just partial information), and when I pointed this out it just kept throwing out different references.

So, it's convenient to be able to chat with your "search engine" to drill down and clarify, etc, but a big time waste if a lot of it is hallucination.

Search vs Chat has anyways really become a difference without a difference since Google now gives you the "AI Overview" (a diving off point into "AI Mode"), or you can just click on "AI Mode" in the first place - which is Gemini.

You're overestimating the mean person's ability to search the web effectively.

The difference is LLMs let you "run Google" on your own data with copy paste. Which you could not do before.

If you're using ChatGPT like you use Google then I agree with you. But IMO comparing ChatGPT to Google means you haven't had the "aha" moment yet.

As a concrete example, a lot of my work these days involves asking ChatGPT to produce me an obscure micro-app to process my custom data. Which it usually does and renders in one shot. This app could not exist before I asked for it. The productivity gains over coding this myself are immense. And the experience is nothing like using Google.

> 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 get rich on gold! You have not demonstrated that there are 10k uses of AI, you've only demonstrated that there are 10k "businesses" interested in making money off of AI. Just like there were 10k+ crypto-currencies... Just like there were 10k+ "uber but for..." apps. Where are these failed gold-rush attempts now?

Investors are currently rewarding the words "AI", so (to extend the analogy) when the gold moved, the gold rushers moved to where they thought the gold would be.

Also your emphasis doesn't change the reading of the sentence.

The problem for him and other leaders in the AI ecosystem is that they ran ahead of themselves to overpromise and that has muddied the water from a consumer perspective. That ultimately increases the risk that we throw the baby out with the bathwater as people get sour with it. I have no doubt LLM will be a feature of society going forward, but they have been reckless with the way they've gone about messaging it

I'll explain it even better, trying to find something useful for the AI to do is the same is trying to find something useful for humans to do. Think that one through. You see, in a room that used to be just you and the boss now includes these whole other new entity called AI. Your boss is really interested in this new thing.

That is all.

The number of startups is no proof they’re doing anything useful. It only proves that the founders think they can make money with it.

> 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.

I don't think you meant "ostensibly".

i found good use for ai but no one cares about my world changing ambitions because they can't see unicorns anymore.

I feel his actual subtext underneath is "We need to do something useful with AI... because I invested in it". And he also goes on saying that calling it slop is bad for society. No, it's bad for the billionaires that invested in it.

You are exactly right!

However, i would say Satya Nadella is quite different from other American CEOs (whose mantra is "greed is good") in that he comes from an Indian middle class family whose focus was on education/good-work and also being forced to take care of a son who suffered from cerebral palsy. All of these shaped his worldviews to be more empathetic of the "common man" which is reflected in his style of leadership.

Similar to my thoughts. If we are still scrambling to find stuff the average Joe finds useful, the 100s of Billions poured into this gold rush are wasted (IMHO).

This looks more like an attempt of gaining scarce electricity.

If a country/state has to choice of giving power to data center A or B, it makes sense for Satya to make statements about how only Microsoft provides the most AI value

Also note that he's not saying Microsoft must find a use for AI, but that customers should.

He's blaming customers that his product isn't hitting the valuation he wants.

Ai and the energy required to power it does partly explain why Trump is so keen to setup American data centres in Saudi Arabia, and why he is so obsessed about Venezuelan oil.

The most statistically likely output given your diligently described symtoms could still be useful. The prohibitive cost in healthcare in general is likely your time with your doctor. If you could "consult" with a dumb LLM beforehand and give the doctor a couple of different venues to look at that they can then shoot down or further explore could likely save time rather than them having to prod you for exhaustive binary tree exploring.

This from a huge LLM skeptic in general. It doesn't have to be right all the time if it in aggregate saves time doctors can spend diagnosing you.

No, doctors are smart enough as a group to have inserted themselves as middlemen and codified it into law, so it will not revolutionize healthcare in a meaningful sense of cutting through the bureaucracy. You may be able to use LLMs to get a suggested diagnosis once tests and symptoms are communicated, but you're going to need to go the doctor to get a referral for the tests/imaging, for formal recognition of your issue (as needed for things like workplace accommodations), and of course for any treatments as well.

At best and if you're lucky to have a receptive doctor you can use it to nudge them in the right direction. But until direct to consumer sales for medical equipment and tests are allowed, the medical profession is well insulated. It is impossible by regulation to "take healthcare into your own hands" even if you want to.

> Evangelists keep insisting that healthcare is one of the things that AI will revolutionize in the coming years, but I just don't get it. To me it's not even clear what they mean by "AI" in this context (and I'm not convinced it's clear to them either).

It's a more-or-less intentional equivocation between different meanings of AI, as you note, machine learning vs generative AI. They want to point at the real but unsexy potential of ML for medical use in order to pump up the perceived value of LLMs. They want to imply to the general public and investors that LLMs are going to cure cancer.

Totally anecdotal, but recently my wife had to go to urgent care for something wrong with her ankle- They send a 4-5 page sheet of arcane terms and diagnoses to her care app (relayed to me via text) and I just slammed that into gemnai and asked "what does this mean" and it did quite well! Gave possible causes, what it meant for her in the long term vs short term, and ways to prevent it. I had a better understanding of what was wrong before the doctor even got to my wife in the waiting room!

Obviously still double check things, but it was moment of clarity I hadn't really had before this. Still needed the doctor and all the experience to diagnose and fix things, but relaying that info back to me is something doctors are only okay at. Try it out! take a summary sheet of a recent visit or incident and feed it in.

Not to mention that the LAST place I want an all-consuming, privacy-destroying data beast is anywhere near my health data.

Respectfully, your positive view of an LLM writing marketing copy is akin to a marketer thinking an LLM codes really well.

It's useful for bribing politicians, milking money from saps, and a bunch of other stuff!

> Recently got a Matic robot

This probably has nothing to do with gen AI (the kind of AI Nadela is speaking about).

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.

Interesting to see people creating throwaway accounts for comments :)

As AI myself I call BS. If I had eyes, I’d roll them into the back of my head so I wouldn’t have to read this idiotic comment

True, ChatGPT and Claude helped me write this sentence.

Don't forget porn.

Though it's a use case people like Satya will want to avoid for reasons.

spam implies low effort BS used in a low-hanging-fruit sense

LLMs will be used for aggressive, yet incredibly subtle manipulation, consensus building, and response tracking.

20-40% of social media is already bots, and in the future it is likely you will not be able to reply to anything anywhere without a bot either 1) responding, or 2) logging and sending your response to multiple parties instantly.

If the Stasi had LLMs the Berlin Wall would have never fallen

> Customer service (chat, phone)

Only because companies have been cutting costs for decades here. This is not a good argument for AI.

> writing software

If you mean typing characters quickly, yes. Otherwise, there’s still a lot of employed devs, with many AI companies hiring.

> writing docs about software

The most useful docs are there because they contain info you cannot determine from the code. AI is not able to do this.

> computer graphics (animation, images)

If you are producing slop, yes.

> driving cars

True, but only because of its improved physical awareness. ie it’s a mechanical gain (better eyes, ears, etc) not an intellectual one (interpreting that information). Self driving cars aren’t LLMs and not really applicable here. Entirely different field.

> AI is only going to become more ubiquitous, there's 0% chance it 'fails' and we return to 2020

Absolutely true. But not for the reasons you think.

> Customer service

An AI might be better than an indian call center but I doubt that when the AI is made by indians anyway.

> writing software, writing docs about software

I have asked AI about exactly one topic and it lied about the API of a library making up the functions I was supposed to call.

> computer graphics (animation, images)

I have indeed seen many wonderful meme images come out of the generators but that was before they got lobotomized for producing that subject matter

[EDIT]

And the worst part is these are all just more "software as a service" designed to remove the possibility of using a tool without approval.

I have a personal budget app and every so often I try and get the latest model to compare my data against the statements and find any discrepancies.

Every time it hallucinates visits to Starbucks.

I never go to Starbucks, it’s just a probable finding given the words in the question.

This should work. I want it to work. But until it can do this correctly all analysis capabilities should be suspect.

Have you tried to get LLMs to do math or quantitative analysis? They're remarkably poor at it

This still reads as "you're using it wrong" to me. Nadella's position is that AI spending would easily justify itself if only the plebs would use it as much as he thinks they should. If only the common man could see the prophetic vision of a coked out tech executive.

My non tech friends/family use AI to ask for silly stuff (they could google it), or just to ask silly questions and see how they react. We have a relative not that famous but maybe known in a niche and they spent like a whole weekendd sending screenshots of GPT, where they asked if this person was known, who was this person, etc.

They don't find AI useful, just a toy. Is their fault? Maybe.

Likewise I keep seeing all these comments on HN about how AI is revolutionary and all these AI skeptics are just haters. I really want to understand what this gap is between the believers and skeptics.

I have access to all the popular AI tools from work for free, I use them for the same cases you mentioned like search, consultation, a better StackOverflow, and autocomplete. It’s definitely useful but I would describe that as incrementally useful, not revolutionary.

Satya is saying that AI needs to start doing more than vibe coding and autocomplete, there’s probably half a trillion invested into the technology worldwide now and it’s not enough for AI to be a good coding assistant. It needs to replace customer support, radiologists, and many other professions to justify the unprecedented level of investment its garnered.

Is your AI faulty? Did you bother asking it for a sentiment analysis of the comments here before drawing your conclusions? That's not what the comments here are saying.

> saved me SO much time I decided to pay for it

AI is subsidized for the users

Energy costs have already risen substantially[0], but the increase has been slower, and it's garnered a bit less media attention than the recent leap in PC hardware prices.

[0]: https://www.bloomberg.com/graphics/2025-ai-data-centers-elec...

But wait! There is more!

Copilot Notepad.

Copilot MS Paint.

Copilot Shoes.

Copilot Ice Cream.

You're plaing game of "control" the wrong way. The whole idea of all these investments and movement was that they do have the control over the AI, not users

Replace idea of AI with idea of calculator. Yes, brain is less engaged, and yes, calculator make you calculate stuff faster

Study. Not studies. And with very limited methodology.

It does take social permission in places like Europe. There's a big pressure against datacenters here that use scarce resources. Like here in Holland the power grid is overloaded and companies need to wait for ages to get a connection.

So this is something that factors in hugely in planning permission. What do we get back for it is a question asked a lot. And datacenters are notoriously bad at providing jobs, during construction yes but in the run phase it's mainly low-value remote hands and security stuff.

[dead]

"Right now, he is imagining that he is just fine. And that a ton of burning leather and metal is fully functioning car."

https://youtu.be/M0S3a32RzEo?t=278

its like the quantum wave function: its both a bubble and not a bubble until you look at it (but no one wants to do that just yet)

GenAI is only useful to bump terrible up to mediocre, so it'd be really stupid to spend time honing one's prompting skills. And as you noticed, so far 96% of the population agrees.

I really hate condascending and arrogant stuff like "Generally speaking people are too stupid to use AI properly at least for awhile". There are plenty of tech illiterate people that are far from being "stupid" and they might not care about or like AI. They just need to send emails to family, share photos and videos and have a video call from time to time. For them, AI is worthless.

But they aren't stupid. You sound like a tech bro.