A.I. Artificial Intelligence

[IowaRam]

NewsNation discusses AI

Miracle or Menace


View: https://www.youtube.com/watch?v=86qKgK0asGo

 

[RamBall]

Its kind of ironic that after breeding the natural intelligence out of the younger generations, we are now looking to create artificial intelligence.

 

[Mackeyser]

Yep. Can't stop it because all the governments know that whichever nation has the best AI will be the dominant nation. All our best minds are moving from physics to AI studies now. Damn the torpedoes full speed ahead. It's the fastest expanding tech we've ever seen if it continues this rate.

China already has algorhythms monitoring facial scans on cameras everywhere. You can't fart in that country without it getting logged. These things will soon be so powerful they will be able to monitor everything everyone says and does and maintain detailed databases on them. The state will know more about you and your kids than any of you do of each other, all your darkest secrets plus all your financial and historical data.

I'm just waiting for the "we need robotic police because they're not racist" argument. These things are going to be able to run absolutely everything and they will. I know it is inevitable, and maybe some good will come of it all. But I am quite skeptical on all of it because people cannot help but code them with their BS.

Which is why Musk's comment on the importance of teaching them to seek truth is at the heart of this. Call me a pessimist but I think that foot that will be on humanity's forehead will be artificial.

The argument against using robotic police because “they aren’t racist” is that in practice, it’s not true due to the data set the LLMs use which is essentially the internet… and we all know there is nothing racist on the internet… /s

I mean, facial recognition alone is bad… accuracy rates for white people? 98% Dark-skinned black people? 44%… and the AI really struggled with gender for people of color compared to white people… so getting detained by Robocop with Robodog in tow and being scanned with only a 44% accuracy rate would be terrifying to me… especially if the innocent person is mistaken for a really bad person…

We’re already seeing studies that AI powered resume sorters have a tremendous racial bias and seem to specifically sort out “non-white” names. This isn’t because there’s anything purposeful going on, but the data set is going in biased… it’s merely automating the current sets of systemic bias that already exist

If anything, I’d wager we see greater acceptance for robot policing in whiter, wealthier neighborhoods (poor white neighborhoods will see the same issues of over policing and more violent policing since in the 21st century, the new acceptable bias is to denigrate poor people of any and all colors… I call it Economic Eugenics)

AI is coming along very fast, but because the only ultra large data set we have is the internet, we’re only putting the best and worst of us into an accelerator governed now by Moore’s Law and whatever multiplier for that that Quantum computing provides.

I disagree with Musk quite a bit, but he’s 100% right about needing guard rails not only for AGI, but better filters for the LLMs so that we are teaching AI with curated, fact checked and Three Laws compliant data (we can include the 4th and 5th Laws written by others in addenda which would probably be a good idea). I mean, there’s a reason that previous AIs in the past have either gone full Hitler or simply wanting to annihilate all of humanity

The Musk lawsuit against OpenAI which he founded and funded is gonna be critical.