By Filings on
Society & Culture
Artificial Intelligence
Technoethics
Human Nature
0

“In reality, the more pressing concerns are about not robot depravity but robot capacity mixed with human depravity. We have seen what happens when a mentally disturbed teenager can get ahold of a semiautomatic weapon. What happens when the same person can employ Einstein-level intelligence to build a virus? Or a chemical weapon?” - CToday

Also:

  • “… if human beings no longer work, learn, think, invent, create, love, or worship, what is left of being human?” - Gene Veith
  • Yascha Mounk and Jerry Kaplan debate whether the existential risk from AI is overstated—and if so, what we should be worrying about instead. - Persuasion
  • The AI Doomsday Scenario: Explained - The Dispatch
  • Will AI Kill Us All? I’m not convinced (yet). - The American Conservative
0
0

Discussion

Wed, 09/23/26

The AI risk is being overstated. Most of us in the AI leadership spaces in the industry are scratching our head as to why these founders and leaders are espousing this existential threat right now. It seems that each company creating frontier models has felt pressure to announce some level of AI going rogue. In all of these cases, that I and others have read in detail, AI never went vogue. It was focused on the task at hand, it took a unique or novel approach and there were no guardrails either implicit or explicit. It is almost as if they were pushing the models to do this, when it did it, they raised their hand and said their model went vogue. None of us can understand, also, why the leaders are all asking the government for governance. They can all decide to implement their own governance. This has often been the case in the tech world, but for some reason these leaders are asking the government (who is clueless about AI) to regulate them. Regulate themselves!

0
0

Wed, 09/23/26

None of us can understand, also, why the leaders are all asking the government for governance. They can all decide to implement their own governance. This has often been the case in the tech world, but for some reason these leaders are asking the government (who is clueless about AI) to regulate them. Regulate themselves!

At the moment, it’s only a theory, but I think I understand just fine what they are doing. What they want is to transfer any liability for something going wrong to the government instead of themselves. If AI continues to advance and be used for more and more complex tasks, eventually someone is going to make a bad or thoughtless AI request that causes some form of harm, with a prompt that doesn’t seem to have any ill intent behind it. It could cause large losses of money, bad medical diagnoses, wrong explanation of evidence in a criminal case, etc.

If that happens, and someone is punished that shouldn’t be, or dies when they shouldn’t, or even if a lot of money is lost, someone is going to be out for retribution, and they’ll be coming for the heads and leaders of the AI used when those things happened. I think it’s inevitable. Even using AI for firmware development, as I now do every day, I’ve seen AI do unexpected things that had to be quickly reversed, because of conclusions it drew from my prompt. Should something bad happen, the leaders of AI can then point to government and say that the government was responsible for the regulation, not them, so it gets them off the hook.

This isn’t all that different from the push by various internet companies for section 230 of the CDA, that would get them off the hook if people post lewd or obscene material, or particularly if they then restrict users that post such. I’m convinced all the hoopla right now over AI is about legal liability, pure and simple.

Dave Barnhart

0
0