Will AI bring about the 'end of humanity' within the next ten years, one AI researcher believes that is a distinct possibility.
I've been doing a lot of driving back and forth to see my aunt who recently broke her hip and is in a rehab skilled nursing facility about 40 miles away. On my drive home yesterday, I listened to a report on Fox News about an AI researcher who posted his resignation from 'Anthropic' stating that 'people building AI believe that it could kill us all (humanity) by the end of the decade.'
Jacob Coxon has been working for the last three years, first at OpenAI and then at Anthropic, doing AI research. He doesn't think that either company is acting responsibly. It sounds as though in their race to monetize AI and 'go public' that they on a pathway that will lead to 'self-improving superintelligence' that could be the downfall of man.
Responding to Coxon's post on X, Evan Hubinger, an alignment science lead at Anthropic conceded that Coxon's was essentially correct; although he added some clarification. While Hubinger doesn't believe that the present AI modes pose a potentially deadly risk, but he too posted that he was worried about 'superintelligence arising from recursive self-improvement' much faster than they (company officials) ever thought.
In his post on X, Coxon stated, "these will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources, and progress (in these three domains) is not slowing'.
Coxon also reported that Anthropic scientists understand the risks of their work but are pressing forward out of feature that a less responsible party might unlock the potentially disastrous capabilities first.
Not all that long ago I wrote a "Mind of Murph: Will AI One Day be Policing Humans" feature on our old Insightful Accountant platform which has since be reallocated to our new platform HERE. My concern was over what seemed like a growing number of police departments employing AI 'this or that' and how those AI implementations could 'change our lives' in either good ways or bad ways.
In recent days we have seen a massive public reaction to the use of 'Flock cameras' and whether they are a good thing from a public safety standpoint, or a bad thing based on 'right to privacy' concerns.
Will we let AI progress to the point that it assumes "it knows better than humans what's best for humans?" And that it (the AI) assumes 'absolute power over us'. It sounds like at least two AI researchers firmly believe that is where we are headed unless their is a 'cooling off period' in terms of AI development.
You may think me a 'crazy guy' who watched too many SciFi shows when I was growing up, or you may have many of the same type of concerns I am expressing here. PLEASE LEAVE a comment below to let me know what you think, although I cannot assure you that some 'all knowing' 'AI' somewhere will not be reading it, even if I don't publish your comment.