AI - LLM Definitions, Capabilities and Ethics

[@matthewfriend59]

How far away do you think we are from AI showing some sort of sentience? Is anyone building the truly massive models' just trying to create that now that you know about?

[@happycomputerguy]

OpenAI has stated it's goal is to create AGI, or artificial general intelligence. Whether AGI would be sentient is debatable, but it is likely it would be. I estimate 5 - 10 years away from sentience, and that is conservative...

[@pigetstuck]

AI tends to have an extreme political bias (not surprising based on its development)

[@happycomputerguy]

Ai has all the biases :) It does not think, it just creates a response that is statistically likely to have been created by an average author from the data it was trained upon, with some additional transformations to correct relationships and such. That is why I lean so hard into saying we really need to be careful about how we use the tech, especially for marketing/sales/etc...

[@pigetstuck]

@@happycomputerguy Agreed. You said it doesn't "think"... how would you define human "thinking"? And how is our thinking different than AI?

[@happycomputerguy]

I can not define human thinking, even with a philosophy education, but I can tell you the way AI creates responses is as similar to human thought as a 1980s video game is to reality.

[@pigetstuck]

@@happycomputerguy Why do you think they are so far apart?

[@happycomputerguy]

@@pigetstuck Human thought emerges from chemical, electrical, physical, and I believe quantum interactions. LLMs produce output from, simplistically, an algorithm. Wildly different in terms of level of complexity and opportunity for emergence of behaviour.