Hacker News new | ask | show | jobs
by welfare 1151 days ago
I believe you have to be more specific what kind of AI you refer to.

Machine Learning algorithms, which include some really straight-forward ones like decision trees, linear regression, support vector machines, etc. are pretty ubiquitous.

If you mean generative AI models (e.g. what powers ChatGPT, Dall-E, etc.) then it's a different story.

1 comments

Yeah, maybe my question was a bit too general. I am not sure I would consider decision trees and the like as 'AI' in any sense. They are just another form of algorithms to me. So you are right: ChatGPT and similar tools are the type of technology I'm asking about. From the few answers I got it seems that there are no general search engines left that have not jumped on the LLM and GPT bandwagon. At least nobody has suggested any sites, only questioned me about my motives for asking this.
> I am not sure I would consider decision trees and the like as 'AI' in any sense.

These have always been considered AI, and I don’t see any reason that would change. E.g. https://aima.cs.berkeley.edu/contents.html

They also have somewhat notoriously long been considered not AI: https://en.m.wikipedia.org/wiki/AI_effect
So if I were to replace a large switch statement with a decision table, then that's AI all of a sudden?
It's artificial and makes intelligent decisions, so yes? Even the switch statement was AI before you replaced it.

The definition of an intelligent decision appears as such: A choice or conclusion made after carefully considering various factors, options, and potential outcomes.

If those conclusions are made ahead of time and baked it, does it make any real difference if nobody can tell from the outside? I think not. It's safe to say there's a lot of intelligent decisions in any software in existence.

Wait, it's all AI?

Always has been.

What OP's probably really trying to say is they don't want to use black box AI of sufficient complexity that nobody alive knows how any of it works. That's the only real hard distinction I guess?

Correct, AI has always been a catch-all for many different algorithms and models for statistical analysis. One definition from the 50's is "the field of study that gives computers the ability to learn without explicitly being programmed.”