Have they considered that a chain of reasoning can actually change the output? Because that is fed back into the input prompt. That’s great for math and logic problems, but I don’t think I’d trust the alignment checks.

I’m behind SEVEN proxies!

Even when consent is informed it can still be fucky. Do you think I want to consent to an arbitration agreement with my employer or a social media platform? Fuck no, but I want a job and interaction so I go where the money/people are. I can’t hunt around for a place that will hire me and also doesn’t have arbitration.

Consent at the barrel of a gun, No matter how well informed, is no consent at all.

This, above any other reason, is why I’m most troubled with AI CSAM. I don’t care what anyone gets off to if no one is harmed, but the fact that real CSAM could be created and be indistinguishable from AI created, is a real harm.

And I instinctively ask, who would bother producing it for real when AI is cheap and harmless? But people produce it for reasons other than money and there are places in this world where a child’s life is probably less valuable than the electricity used to create images.

I fundamentally think AI should be completely uncensored. Because I think censorship limits and harms uses for it that might otherwise be good. I think if 12 year old me could’ve had an AI show me where the clitoris is on a girl or what the fuck a hymen looks like, or answer questions about my own body, I think I would’ve had a lot less confusion and uncertainty in my burgeoning sexuality. Maybe I’d have had less curiosity about what my classmates looked like under their clothes, leading to questionable decisions on my part.

I can find a million arguments why AI shouldn’t be censored. Like, do you know ChatGPT can be convinced to describe vaginal and oral sex in a romantic fiction is fine, but if it’s anal sex, it has a much higher refusal rate? Is that subtle anti-gay encoding in the training data? It also struggles with polyamory when it’s two men and a woman but less when it’s two women and a man. What’s the long-term impact when these biases are built into everyday tools? These are concerns I consider all the time.

But at the end of the day, the idea that there are children out there being abused and consumed and no one will even look for them because “it’s probably just AI” isn’t something I can bear no matter how firm my convictions are about uncensored AI. It’s something I struggle to reconcile.

Note that these studies aren’t suggesting that heavy ChatGPT usage directly causes loneliness. Rather, it suggests that lonely people are more likely to seek emotional bonds with bots

The important question here is: do lonely people seek out interaction with AI or does AI create lonely people? The article clearly acknowledges this and then treats the latter like the likely conclusion. It definitely merits greater study.

Yeah but I can just ignore the bullets because they are nerf. And I have my own nerf guns as well.

I mean at some point any analogy fails, but AI is nothing like a gun.

Seems to me libel would require AI to have credibility, which it does not.

It’s a tool. Like most useful tools it can do harmful things. We know almost nothing about the provenance of this output. It could have been poisoned either accidentally or deliberately.

But above all, the problem is ignorant people believing the output of AI is truth. It’s pretty good at some things, but the more esoteric the knowledge, the less reliable it is. It’s best to treat AI as a storyteller. Yeah there are a lot of facts in there but when they don’t serve the story they can be embellished. I don’t see the harm in just acknowledging that and moving on.

It’s AI. There’s nothing to delete but the erroneous response. There is no database of facts to edit. It doesn’t know fact from fiction, and the response is also very much skewed by the context of the query. I could easily get it to say the same about nearly any random name just by asking it about a bunch of family murders and then asking about a name it doesn’t recognize. It is more likely to assume that person is in the same category as the others and if the one or more of the names have any association (real or fictional) with murder.