← Back to context

Comment by sanskarix

3 days ago

[dead]

Good point. Experience teaches us that the breakthroughs in AI will lead to something, we just don’t know what that something is yet and there is a lot of (maybe too much) speculative betting on what it could be.

> most people hit a wall figuring out what it's actually good for beyond parlor tricks.

That's what my parents thought about computers and the internet, wondering what it's actually good for beyond burning $9000 in phone bills to Zerg rush Protoss noobs.

And all the other things computers+internet could do, they could already do through other more reliable (at the time) ways.

But then it turned out that simply making mundane tasks just a little bit faster, and reducing the need to interact with strangers by just that little bit, created a new step on the staircase, a new baseline, with which to reach and do other grander things more easily.

  • Email was the killer app I think that showed everyone how damn useful the internet is. Then Hotmail showed how convenient email can be.

    What is the AI version of that? Maybe code generation. Maybe.

    • Code generation/LLMs are the SMTP and other internet protocols, we don't have killer apps yet.

    • > What is the AI version of that?

      Being able to plan a trip from a single sentence would be one killer app for many people:

      "I'm free next week. I'd like to go to A, B, or C for a couple days. What's a cheap flight and a room within this budget near X area?"

      and if it could go and also make a booking through your accounts that would be amazing.

      But right now even Google's Gemini is an utter useless dumbass if asked to search Google Flights or Airbnb.

      I mean if LLMs could just be a natural-language wrapper around existing tools, that'd be amazing in itself. But corporivalry has made that an stillborn dream.