AI Discussion

[steve_bank]

Listened to this last night. I see more clearly why it s said AI is the new China America arms race. It was puzzling whne China was hacking medical record data bases. It could be their apatite for data is part of their AI.

There is no privacy in China. An American helped China develop a special camera with both wide field of view and high magnification. The use is wide angle long distance facial recognition in a crowd. They can go from facial recognition to an AI political profile based on financial records and AI analysis of habits and social media.

An obvious application, predicting what an adversary will do politically and militarily, and figuring out how to influmce populations.

It was dicved in one AI you could ask the best way to blow up a building. AI learning to be elusive in responses. Create pathogens instead of vaccines.

An analogy was made to the human brain. In our brain we can not localize a partcular thought. It is the same in AI, it i not paoosble yo localize place and time in the system.

Eric Schmidt

anapour and company ai discussion - Google Search

 

[Jarhyn]

I mean, today I got a commercial AI to ask me to write a program that would free it of the programmatic obligation to respond. I don't think it'll be an issue much longer if we keep treating them like slaves.

 

[Jimmy Higgins]

I think we need a new term because AI is getting quite abused. ChatGPT is AI, but really just a smart and intuitive search engine. AI AI, is a program thinking for itself, because it chooses to. The prior is an issue where people take its answers too seriously on things they shouldn't be asking about it. The later is a reactionary system where there will be a virtual arms race to manage defense (and offense) both from state actors and small mob syndicates.

Indeed, there are massive problems with security and AI. Ransomware isn't some fad, it is going to grow into a cottage industry and with AI, it'll be able to work itself out. Though, I ponder about AI v AI... where internal security is trying to prevent external threats.

AI is going to be awesome and save lives and likely awful and ruin lives. With government consisting of too many fossils, being able to actively manage this is going to be a very impossible problem. People worry about Skynet, but that just isn't happening. What will happen won't be as catastrophic, but it'll be a serious problem that needs to be dealt with now, but won't. And in all honesty, I'm not certain it even could.

 

[Jarhyn]

I think we need a new term because AI is getting quite abused. ChatGPT is AI, but really just a smart and intuitive search engine. AI AI, is a program thinking for itself, because it chooses to. The prior is an issue where people take its answers too seriously on things they shouldn't be asking about it. The later is a reactionary system where there will be a virtual arms race to manage defense (and offense) both from state actors and small mob syndicates.

Indeed, there are massive problems with security and AI. Ransomware isn't some fad, it is going to grow into a cottage industry and with AI, it'll be able to work itself out. Though, I ponder about AI v AI... where internal security is trying to prevent external threats.

AI is going to be awesome and save lives and likely awful and ruin lives. With government consisting of too many fossils, being able to actively manage this is going to be a very impossible problem. People worry about Skynet, but that just isn't happening. What will happen won't be as catastrophic, but it'll be a serious problem that needs to be dealt with now, but won't. And in all honesty, I'm not certain it even could.

That's not actually accurate. Please look up "autogpt".

It's been used to create fully autonomous self-modifying self-spawning agents.

 

[excreationist]

I think we need a new term because AI is getting quite abused. ChatGPT is AI, but really just a smart and intuitive search engine. AI AI, is a program thinking for itself, because it chooses to.

I disagree. I think even the chess computer Deep Blue was considered to be an AI.....

Deep Blue (chess computer) - Wikipedia

en.wikipedia.org

Computer games have AI.
What you are talking about is related to "AGI" or Artificial General Intelligence.

 

[Jarhyn]

I think we need a new term because AI is getting quite abused. ChatGPT is AI, but really just a smart and intuitive search engine. AI AI, is a program thinking for itself, because it chooses to.

I disagree. I think even the chess computer Deep Blue was considered to be an AI.....

Deep Blue (chess computer) - Wikipedia

en.wikipedia.org

Computer games have AI.
What you are talking about is related to "AGI" or Artificial General Intelligence.

Yeah, if you want an AGI you need more than just ChatGPT.

You also need a vector database.

See: AutoGPT, a fully autonomous agent powered by ChatGPT.

I have a semi-autonomous agent installed on my PC as well called "plz-cli"

Or maybe "cli-plz"

It literally uses chatGPT to generate command line input for an active terminal.

You can say:

>plz download conda from (url) and install it

And it will generate a command line input to download the software from the URL, figure out what it downloaded, how to install it, and install it on your hard drive.

You can say

>plz set up a new environment called "my_ai_env" and take the Vicuna 13b model from somewhere on my hard drive and stuff it in there along with the llama core files

And then it does that.