Satya Nadella's Top GPT-5 Prompts

5 pykello 9 8/29/2025, 3:58:01 AM twitter.com ↗

Comments (9)

mrcsharp · 13h ago
Is "How to be a decent human being and an innovative CEO" one of them?

This reeks of desperation.

fennec-posix · 13h ago
Yeah, at the very least extremely tone-deaf.
aurareturn · 13h ago

  This reeks of desperation.
Nah, I think this is the future. You may not like it because HN is very privacy focused and hate giving data away to companies. However, I think it's inevitable that people will want to feed all their work data (and probably personal as well) to an LLM and let it help. I think Nadella's examples are pretty spot on for what current LLMs can do for workers if the LLMs have access to work data.

I think it's exciting personally. I think companies who care about data privacy will run some kind of private inference cloud or private DGX racks.

al_borland · 12h ago
All those prompts seem like they’d have a high probability of hallucinations. Is he making decisions to steer a massive company based on stuff the AI may have simply made up? If a person feeds me bad data as often a Copilot does, I wouldn’t trust them with anything.

It’s incredible the error rate we are willing to tolerate from AI. It’s drastically higher than the error rate we would ever tolerate from a coworker.

Considering how much money Microsoft put into OpenAI, he has a very vested interest in hyping it to try and get it to pay off, but I can’t believe he actually relies on those queries.

mrcsharp · 13h ago
> You may not like it because HN is very privacy focused and hate giving data away to companies.

This is a big assumption from your side. Care to back it up? Where did I talk about Privacy?

aurareturn · 13h ago
Apologies. Not referring to you specifically. Just general HN opinion. People here values data privacy far more than the general public.
mrcsharp · 12h ago
I didn't mean to sound aggressive, apologies for that as well.

Let me explain: in my view, he is doing a lot to try and hype up Microsoft's AI offerings and this latest attempt feels like the kind of thing an employee does to hit some arbitrary KPIs.

aurareturn · 12h ago
He's the CEO of Microsoft. I don't blame him for trying to hype up Microsoft's AI offerings.

But at the same time, I like what he demonstrated even if it doesn't work as well as his demos. I believe all work data should be ingestible by an LLM. I'm just anticipating a wave of opposition on HN because people here value data privacy far more than the general public.

starmole · 13h ago
Twitter link is unreadable by clicking on it.