We urge policymakers to instead draw on solid scholarship that investigates the harms and risks of AI—and the harms caused by delegating authority to automated systems, which include the unregulated accumulation of data and computing power, climate costs of model training and inference, damage to the welfare state and the disempowerment of the poor, as well as the intensification of policing against Black and Indigenous families. Solid research in this domain—including social science and theory building—and solid policy based on that research will keep the focus on the people hurt by this technology.
YES
I don’t think Ai Is going to go skynet. I know it’s going to be used to disenfranchise Black people, destroy creative fields, and generate mis- and disinformation because it’s already doing it
This is such an important distinction. Current AI is incapable of wanting to cause any of that harm, yet it’s already happening. The danger won’t be skynet, it will be and always has been human greed and ignorance.
Advocacy groups like the EFF and ACLU have been raising the alarm about the surveillance state for decades. It has very little to do with AI, and regulating AI will not likely have any effect on state-level abuse, because guess what? They already operate above the law and beyond reason. That’s the real problem.
We urge policymakers to instead draw on solid scholarship that investigates the harms and risks of AI—and the harms caused by delegating authority to automated systems, which include […] damage to the welfare state and the disempowerment of the poor […]
I got an example of that yersterday, when calling FedEx’s over the phone to schedule a pickup.
Contrary to calls I’ve made in the past, it was answered by a chatbot instead of humans. This raise various problems, in particular for already-disenfranchised people who rely on customer service being available over the phone.
- A disclaimer says “This call may be recorded for training purpose”, there was no asking for consent, no way to opt-out. Does that mean my voice and conversation will be fed into a database to develop DL/LLM AI systems? There’s no simple way to know, or to refuse without hanging up (ie opt-out = being denied service).
- The chatbot had a hard time understanding my inquiry, forcing me to repeat/rephrase it multiples times. In the end it replied it’s not possible to schedule pickups over the phone. This was possible back when humans were on the phone. Internet access is now required. Meaning disenfranchised people with no (easy/dependable) Internet are being left out.
- The chatbot used a generic greeting, didn’t introduce itself properly, and was evasive when I asked “are you a human?”. Not being forthcoming will confuse some old or non-tech-savy people who won’t easily realize it’s a robot, not a human.
I don’t know, I’m much more concerned about the possibility that we develop huge automation capabilities that end up being controlled by very few people.
As for the specific issues in the article - yes, they’re real problems. But every advance in communication and information technology makes it easier to surveil or defame, and can be used for bad policing.
Right now there’s a push to regulate the internet to “prevent CSAM” by blocking encryption, and I’m afraid a push to regulate AI will not get better results.
Sure, we can ban predictive policing and demands some amounts of transparency (and the EU already wants to do that). But if we try to go further and impose restrictions on the AI models themselves, this will most likely solidify that AI is controlled by few powerful corporations. After all, highly regulated models by definition can’t be free and open.
But every advance in communication and information technology makes it easier to surveil or defame, and can be used for bad policing.
So we should just give up and accept surveillance and defamation without trying to come up with any legal counter measures?
Right now there’s a push to regulate the internet to “prevent CSAM” by blocking encryption, and I’m afraid a push to regulate AI will not get better results.
Totally agree that KOSA and the like are awful, but the existence of shitty regulations doesn’t negate the need for positive regulations.
But if we try to go further and impose restrictions on the AI models themselves, this will most likely solidify that AI is controlled by few powerful corporations. After all, highly regulated models by definition can’t be free and open.
I just don’t see this? So regulating powerful companies use of AI will… solidify their power? I’m not connecting the dots here but that might be on me, as I think there are plenty of highly regulated spaces that still see innovation.
Your assessment seems spot on to me. I’m connecting some projected dots to late stage capitalism. Perhaps the AIs will trickle down and such if we hold off on regulations.
Of course it’s possible for the government to impose regulations without sticking their face in and motorboating the AI’s contents. Google, Microsoft et al. would love to prevent this from happening because they actually do have their faces in there.
Some regulation proposals seem fine to me, like the proposed EU AI act.
But for some of the problems the article lists, like defamation or porn generation, you just can’t prevent if you have free and open models out there. You can make these things harder - and people already work on that - but if I have a free and open model, I can also change it (and remove restrictions).
The only way to stop those uses would be to keep AI tightly controlled in a walled garden. In capitalism, those walled gardens will belong to companies.
I think their examples are the kind of actual outcomes that should be addressed. Yours absolutely also are, but their core concept is “focus on the actual shit we see being fucked up instead of imaginary future shit that’s might theoretically eventually be a concern but is just not a real world issue with anything resembling the tech we have today”.
Didn’t read the article but headlines like this are extremely naive. If a person takes seriously the existential risk factor of AGI/ASI then that’s by definition the most important thing one can pay attention to. That’s like asking a deeply religious person to stop worrying about hell.
But to build on your analogy: we don’t make regulations based on a religious doctrine anymore in most countries. If your religion says no one is allowed to wear mixed fabrics or eat pork, that’s fine if you’re not doing that, but we’re not banning those things for all of society.
There’s about as much proof for an existential AGI threat as there is for a deity, so let’s not make policies based on either, and focus instead on real potential and already proven harms of AI
I have a hard time taking anyone talking about AI regulations seriously. Unregulated AI is already out there and you don’t get the ghost back into the bottle, it’s already working beyond all regulations and laws. Deep fake porn and the environmental cost, both seam to be things we need to accept, I don’t see them being worse than Crypto mining energy waste (which thankfully is slowing down) or already existing porn on the internet. Deep fakes make it even possible to have a deniability claim, once your real porn got out there, which is a positive thing as you could always claim it to be fake, reducing the possibility of blackmailing.
AI art, music are the things that will be replaced to large extend, as the results we have already are really good. Now human art will still be desired by some, but it will become rare and people will do this as hobby from now on. Thing is, good AI art as of now still needs a bit of a technical understanding and time generating, much less than before, but I don’t think this can fully go away, even if you add the names of an artist to a prompt. You need to understand a composition and the technical background.
Hard to predict the future, but creating something unique, the AI doesn’t fully know yet, is still challenging. Stuff like Stock photos, (thankfully) will be replaced, as anyone can just add prompts to generate generic images, and in future videos.
The writing part is the most concerning, as most nations lack an independent news outlet. For example in Germany we have the Rundfunkbeitrag, which pays for state independent news. It’s pretty resistant to fake news and as neutral as it gets, compared to the commercial news out there, who’s goal it is to generate outrage to sell clicks.
No matter what, there’s going to be a time, where too many people are not up to date yet, but we’ll also have enough people able to manipulate context via AI. People will need time to adjust. Only after this time of unrest most people will understand what’s true and important. Maybe it will have unintended effects we don’t know yet. What if conspiracy completely dies out as anyone can generate conspiracy text or anti-conspiracy text, making them all look like fools? What if we need more people fact checking and image/videos for error checking? What if we need more people who can understand the difference between a generic script and a good one? It’s like with machines, people used to do all the stuff by hand, now a machine does it, but we need more people who fix and control the machines. Yes it’s on a whole new level, the unemployment rate might be hit the hardest by this. The more we automatize stuff, the more we need to talk about UBC, unconditional basic income.
