Quick AI Summary of the Cuse Post Game Thread

[businesslawyer]

That's pretty cool. Can we get a summary of the Key Tweets thread next?

I can see the story now. Nerdy thread on UConn bulletin board sets AI back 100 years.

 

[B1GEast]

I meant that AI (at least as it exists now) is not like what we think of as “AI” shown in fictional media. It can do a lot but it isn’t “thinking” like a person. It’s just analyzing data. It doesn’t understand the intangibles of sports, it can only look at the statistical analytics.

That said my biggest issue with AI is — beyond that it’s a giant drain on natural resources and the energy grid — that it’s become a tool for laziness. I don’t need an AI to tell me why UConn lost the game. I know why. And if I wanted an analysis of the finer points I would simply look to the insights of people with better football knowledge than myself.

That's the real problem, cognitive drift and dissonance. AI and social media is making us dumber over time.

The more you depend on intelligence to do things, your brain is actually being rewired.

Pattern framing may not be reversible for some people.

It's a real thing. AI can help do tremendous things. In healthcare it's already improving diagnosis and helping with genetic sequencing and so much more.

There's a lot of good, but also a lot of bad since most people are glued to their phones.

You need to do things that stimulate the brain, reading, exercising and meditating, etc.

 

[ShakyTheMohel]

 

[B1GEast]

It's programmed to provide a result. It returns a token. GPT has tens of billions of parameters if not a lot more.

There probably nothing that it hasn't seen since it uses not only the LLM that it's trained on, which is exhaustive, but it also uses RAG.

Mark is a smart guy, but his statement is off by a mile.

It's not about returning a response with I don't know as much as it is I don't understand contextually how to provide an answer.

I've seen the transformer respond with "here's where it gets tricky" and indicate that the response may not be conclusive and then say if you provide more detail so that I can better understand.

That logic that I described is incredible.

And yes, it does hallucinate at times, like really fail, and there can be many reasons for that, like poor prompting or overwhelming it with too much data.

GPT is at a PhD level and for most of what I use it for, I consider it to be an incredible force multiplier.

We have only scratched the surface, it's scary.

 

[ShakyTheMohel]

It's programmed to provide a result. Yo return a token. GPT has tens of billions of parameters if not a lot more.

There probably nothing that it hasn't seen since it uses not only the LLM that it's trained on, which is exhaustive, but it also uses RAG.

Mark is a smart guy, but his statement is off by a mile.

It's not about returning a response with I don't know as much as it is I don't understand contextually how to provide an answer.

I've seen the transformer respond with "here's where it gets tricky" and indicate that the response may not be conclusive and then say if you provide more detail so that I can better understand.

That logic that I described is incredible.

And yes, it does hallucinate at times, like really fail, and there can be many reasons for that, like poor prompting or overwhelming it with too much data.

GPT is at a PhD level and for most of what I use it for, I consider it to be an incredible force multiplier.

We have only scratched the surface, it's scary.

I have seen it hedge. I haven’t seen it say “I don’t know”.

I am amazed and impressed with what AI can do, but I do think Cuban’s tweet is valid.

 

[B1GEast]

I have seen it hedge. I haven’t seen it say “I don’t know”.

I am amazed and impressed with what AI can do, but I do think Cuban’s tweet is valid.

When people say I don't know it's because they recognize there's a gap in understanding.

For a machine that spits out probability, that's not always tied to comprehension.

He's talking about self awareness when the AI is simply making a best guess.

Cuban''s remarks frame this weaknesses as an inability to admit ignorance when it's just a machine parsing a question. He's off base

What is he going to say next that it wasn't sorry for not providing a rich enough response?

Humans don't know, machines don't compute.. There's a huge fundamental difference there.

 

[HCCForever]

I have seen it hedge. I haven’t seen it say “I don’t know”.

I am amazed and impressed with what AI can do, but I do think Cuban’s tweet is valid.

I’ve worked with many people who never admitted to not knowing. That viewed it as a weakness instead of taking the opportunity to learn what they didn’t know.

 

[ShakyTheMohel]

I’ve worked with many people who never admitted to not knowing. That viewed it as a weakness instead of taking the opportunity to learn what they didn’t know.

Exactly…and that’s why AI not being able to say “I don’t know” is a bad thing.

Basically you are saying AI is as good as the worst people…which says that right now it’s not the be all end all…..yet.

 

[FfldCntyFan]

I can see the story now. Nerdy thread on UConn bulletin board sets AI back 100 years.

Wouldn't that be a good thing? Personally I'm in no hurry to see Skynet enslave us all.