In a profession covering several decades, artificial intelligence researcher and professor Stuart Russell has contributed extensive lore on the subject, including foundational textbooks. He participated us onstage at TC Conference: Robotics+ AI to discuss the threat he comprehends from AI, and his work, which proposes a tale solution.
Russell’s thesis, which he develops in” Human Compatible: Artificial Intelligence and the Problem of Control ,” is that the field of AI has been developed on the fictitiou premise that we can successfully define the goals toward which these systems work, and the result is that the more powerful they are, the worse they are allowed to. No one really speculates a paperclip-making AI will expend the Earth to maximize production, but a crime-prevention algorithm could quite easily take seriously constructed data and objectives and turn them into recommendations that movement very harm.
The solution, Russell suggests, is to produce arrangements that aren’t so sure of themselves — basically, knowing what they don’t or can’t know and would be interested to humans to find out.
The interview has been delicately edited. My remarks, though mainly insignificant, are retained for context.
TechCrunch: Well, thanks for joining us here today. You’ve written a book. Congratulations on it. In fact, you’ve actually, you’ve been an AI researcher and columnist, professor for a long time. You’ve seen this, the field of AI sort of graduated from a niche orbit that academics were working in to a global priority in private industry. But I was a little surprised by the thesis of your diary; do you really think that the current approach to AI is sort of basically mistaken?
Stuart Russell: So let me take you back a little, to even before I started doing AI. So, Alan Turing, who, as you all know, is the father of computer science — that’s why we’re here — he wrote a very famous newspaper in 1950 announced ” Computing Machinery and Mind ,” that’s where the Turing test comes from. He laid out a lot of different subfields of AI, he proposed that we would need to use machine learning to create sufficiently rational platforms.
Read more: feedproxy.google.com