Technology

On the negatives:

• Online abuse: AI is fueling the spread of sexual abuse material and sexually explicit deepfakes, with women and children most at risk.

Eh. I don't see this as that fundamental. I mean, Photoshop lowered the bar too to making synthetic pornography, to slapping a head on someone else's body. The ability to do image warping allows resizing body parts in a convincing-at-first-glance look, and I remember when that was a fad. It just doesn't seem to have changed society all that much in the past. It may be that people just stop caring much about pornographic images of a particular person. I'm not saying that it won't have an impact, but I have a hard time seeing a scenario where it really deeply alters society.

• Disinformation: AI can generate false information that is as convincing as the truth, undermining trust in public debate and democracy.

Yeah, that's a bigger issue.

It is not one that is impossible to deal with. It was a situation that we had to deal with prior to recording technology. And there were problems

like accusing politicians of going to one place and saying one thing in their speeches, and then to another and saying something else was a real thing in the US back when the only record we have was from newspapers printing summaries of what they said. We had solutions for that era, like people who would put their reputation on the line to attest to various facts. We could do it again. But we've benefited from having easy technology that made it pretty easy to make a credible record cheaply and easily of all sorts of things, audio recording, photographs, and video recording, and the bar for that might rise.

It was always going to happen one day, whether-or-not neural networks were involved. Computer graphics and audio synthesis have only gotten more-accessible and convincing over the years.

I think that one new issue is the ability to synthesize propaganda using the bandwagon effect on social media. That is, it used to be just financially impractical to have a zillion people online trying to influence people. But...if chatbots drastically lower the cost to that, that could be a real issue, and one that we haven't had to deal with before, ever.

Crime: Criminals are using AI to carry out cyberattacks, fraud and social engineering scams.

Yup. Some of this we can fix, and is because we have outdated authentication mechanisms (like recognizing someone's voice on a phone because the phone network has basically no authentication mechanism). Some of it is going to be harder. I think that this is largely not fundamental problems, but some things are going to have to change.

• Mental health: Some AI systems can reinforce harmful beliefs or behaviours, leading to mental health crises, including suicide.

Ehhh. I mean, yes, but so can all sorts of other things. Sitting alone watching TV all the time. I remember distress when the Internet became more widely-available about how people could enter into harmful forums. And we've always had crackpots with weird ideas out there.

https://en.wikipedia.org/wiki/Time_Cube

Ray didn't need AI to develop his theories.

Honestly, I'm kinda more concerned about stuff like cults and scams, cases where someone is actively attempting to maliciously manipulate people than "random person goes and spends time talking to a chatbot and feels that it reinforces his crackpot views".

• Loss of control: As AI becomes more autonomous, experts warn it could become harder to monitor and govern without stronger safeguards.

Yeah, this is a big one. The Friendly AI problem is a hard one. I don't know if there are practical solutions, not for self-improving advanced intelligences, which we are certainly going to try to build.

• Environmental impact: The energy-hungry data centres which power AI are contributing to greenhouse gas emissions which leads to global warming.

Eh, yeah, but this is basically the same as our existing energy problems. Like, we were already emitting an unsustainable level of carbon dioxide emissions. The answer has to be shifting to different generation methods. The answer was never going to be "we keep burning coal and whatnot and everyone just reduces energy usage enough on a per capita basis to keep human-driven carbon dioxide emissions at a sustainable level".