Joe Biden Wants US Government Algorithms Tested for Potential Harm Against Citizens

jeffw@lemmy.world to Technology@lemmy.world – 738 points –
Joe Biden Wants US Government Algorithms Tested for Potential Harm Against Citizens
wired.com

Highlights: The White House issued draft rules today that would require federal agencies to evaluate and constantly monitor algorithms used in health care, law enforcement, and housing for potential discrimination or other harmful effects on human rights.

Once in effect, the rules could force changes in US government activity dependent on AI, such as the FBI’s use of face recognition technology, which has been criticized for not taking steps called for by Congress to protect civil liberties. The new rules would require government agencies to assess existing algorithms by August 2024 and stop using any that don’t comply.

124

You are viewing a single comment

Regardless LLMs aint the most revolutionary tech of the century

Oh it is though. Not LLMs, they'll probably go out of date soon. AI is the thing that's revolutionary. Only second to genetic modification, but AI and most other forms of human advancement will likely be hand in hand.

Like solar roadways, or the hyperloop, or the thorium powered car.

This is irrelevant but I didn't buy the crap on any of those.

Or crypto.

AI is actually producing results. It's in use as we speak. It's producing real value. It's not some fantasy far off thing. You can use them today.

And you probably use AI every single day when you open your phone and take a picture. You'll eat these words, and when you do you'll hopefully have access to the tech yourself, because otherwise Mister Microsoft will control who can and can't succeed in the world.

Dude im using a beat up s7 so in lucky if the camera wants to work, im sure as shit aint getting ai on it and its updateless crap.

And secondly I really want nothing to do with "AI" (fucking hate that marketing labeled it that) and I kinda hope it all falls in on itself longterm.

Short of it working as generally software improvements or added tools its looking to be at best a new tool for corporations to exploit us with. Hyper targeted ads and all that. Frankly I severly doubt its gonna lead to much more than some slightly imporved automation for manufacturing. It will probably crash face first into a wall once computer tech starts hitting the power limitations for computation again.

Short of biotech allowing us to combine organic brains and computers I suspect we're already approaching the "AIs" limit.

Also can we stop calling it fucking "AI" at best its simulated intellegence theres a fucking difference ones sapient or maybe even sentient and the other is just a really smart calculator

im using a beat up s7

You understand AI has been a thing since like 2013, right?

generally software improvements or added tools its looking to be at best a new tool for corporations to exploit us with

Which is why we need this stuff to be open source

Whats your definition of AI here, cause I legit can't tell at this point. Are we talking algorithmic learning or are we just talking basic non-learning software that people call AI but is basically just a general improvement on say stabilizing video in a recording?

AI is a layname the field of deep learning. Systems that use data to create a function that performs any task.

There are other branches of machine learning but deep learning with neural nets blows them out of the water when it comes to "pushing the envelope", and aren't really what people think of when they think of AI- even if most use cases really don't need it.

I'd be against regulating those algorithms as well, but they're so boring nobody really cares and they'll always be available open source.

basic non-learning software that people call AI but is basically just a general improvement on say stabilizing video in a recording?

Even chat GPT and other LLMs are not learning as you use them. A really big and growing fraction of stuff your computer does depends more and more on trained neural nets.

Im aware of the fact their trained on neural net and yada yada, im just fucken tired and have hit the shit at words stage.

Also I realized that most of the focus for these laws and whatnot are probably on advertisement AI or atleast use in that way and maybe self driving cars. Training an AI to do open heart surgery probably aint on the chopping block.

Which makes me wonder if I should've opened with that point in my initial comment. Regardless I do think you're overreacting a bit on the political front. Frankly slowing R&D down a bit can be good a lot of folks rush shit and it redults in damages due to immature tech. As bad as the dems can be remember the best of the republicans are allied with technofeudal libertarians and christofascists. Dont let perfection be the enemy of good enough, if the price to pay for long term stability is slower R&D so be it. Rome lasted 2000 years by being slow and meticulous, Alexander the greats empire lasted a whole 5 minutes after his death cause he rushed the whole damned process.

Now im gonna go get food from taco bell before I get philisophical and insufferable, I like being a redneck bezerker not a fucking SoCal Socrates.

Regardless I do think you’re overreacting a bit on the political front

Biden hasn't crossed the line yet. I'm not reacting at all, just cautious and making my opinion clear as a way to (ever so slightly) encourage further regulation not to happen.

Dont let perfection be the enemy of good enough, if the price to pay for long term stability is slower R&D so be it

It's not slowing development. They're making sure the common person doesn't have access to it. They are going to ensure only big companies see the benefit.