3 ms·
"There can be no question that Ai must be regulated and controlled just as powerfully as we control money." That's the single new idea in there. It might be a
by Animats 2y ago
"There can be no question that Ai must be regulated and controlled just as powerfully as we control money."
That's the single new idea in there. It might be a good one. Or not. But it's worth thinking about.
AIs that you can only talk to have some risks, but most of the risk is that they say something that embarrasses or annoys something, or that they are believed to be right when they aren't. That's about the level of risk Youtubers generate. AIs that can spend money, buy, sell, and use the power of money - they can take over. If they're good enough at capitalism.
- klabb3 2y agoAgreed. To elaborate, he talks about self-identification, that anything with AI should be required to be labeled as such. And that enforcement is similar to how globally governments chase down counterfeiters of money. It’s a good reference, in times where regulation is weak and easily circumvented – that hard lines and perseverance to maintain control at a global scale is still possible. Weapons and export controls are somewhat in a similar category. You still can’t build recreational nukes. Still, would that actually happen? As usual with safety, rules are (only) written in blood. To most politicians, I suspect these ideas are indistinguishable from the constant chatter of doomsday cults and alarmists. Without a major disaster resulting in loss of life, war, famine, I doubt anything’s gonna happen. Even a global IT outage is forgotten after 2 weeks, and after Covid we’re pretty desensitized.