Thursday, October 27, 2016

Life threatening risks from misuse of AI, terminators, etc.

One of the folks on an expert list specializing on computational intelligence posted:
Paul’s comment resonates well with my personal view against the popular (non-expert) view of E. Musk, S. Hawking, etc… that soon we need to worry about the AI overlords!!! (by watching too many Terminator sequels!)

Good morning,...!

I am sorry if I gave a one-sided picture, and did not say enough about why I worry even more than Musk and Hawking do about misuse of AI and of IT in general. Having gotten very deep into some of these issues, I worry more and more about the level of probability that the human species will become extinct well before its time (before 10,000 AD or much sooner). Misuse of IT is on my short list of grave and urgent concerns which could lead to extinction, though it is not at the top of the list; misuse of nuclear technology and little-known dimensions of climate change are. (Credible friends also  warn me not to underestimate potential for fatal misuse of some bio technologies.)

Obama has suggested a US government policy which is basically "wait and see":

But competing visions of IT are already fighting it out both in the market and (more seriously) in heavy-handed political spheres which can short-circuit the market. Some of the greatest risks come from overuse and overempowerment of technologies which pretend to be intelligent but are not (if we use the word "intelligence" in the traditional way, unlike PR salesmen who label almost anything narcissistic as "smart"). The situation is so grim that I went through a lot of groping and soul searching when they asked me to give the keynote talk at the IEEE CIS Winter School on Big Data in Computational Intelligence; "what can be said of a positive nature to the next generation? Is there ANY active path forward which does not make life worse and riskier in the end?" There is, but it requires more intelligence and will than I see in any of the political or corporate forces now active in that space; I posted links to the abstract and the talk ("under "added in 2016" near the top). 

NATO recently pressed me for more detail, and so I have a book chapter in process on some of these issues in the NATO series, posted at

I have run across many groups of people vigorously feeding into all three transition teams -- Clinton, Trump
and... imperial or Trajan (planning for the transition after they get rid of Clinton). It's an important dialogue, though I find it a bit overwhelming in some ways; my response this AM to one thread of that:

I have to admit that some of these posts have strengthened the emotional response which says that we are going to hell regardless of who gets elected. Earlier, it seemed (at first, on a superficial read) someone posted: "Now let us be truly progressive. Let us go from 'a chicken in every pot to a chip in every brain." Knowing a bit about chips and about brains, my emotional response was: "If that's progressive, maybe we should think twice about the possibility of instead electing someone who  would only give us a rerun of the Great Depression of the 30's  and the resulting world war (this time nuclear)."  

But ..'s position on the federal reserve is actually much more clearly aligned with the great and complex historical hermeneutics of Trump and of the Koch movement -- not the Clinton transition planning at all. Lots of nice words, but no real understanding of how money supply and interest rates work, even to a first approximation. 

For a quick first approximation see:

(Having built econometric models which actually worked on real data, better than any of their predecessors in the sectors I studied, I do know at least the basics, but 
full competence in financial policy requires another level beyond that.)

Trump has argued that present policies and structures are getting us inexorably and slowly to hell. It seems he is right. But his clear and definite formulation of monetary policy makes it clear that he would simply get us to hell more quickly and more definitely. (Reminds me of the old adage about the road to hell paved with good intentions. No substitute for hard work to understand how things really work.)

But... time for me to focus on what I personally should be doing today, which is off in the realm of quantum optics and such. No more hermeneutics.

But no, no shrinking from the responsibility to vote, to make a judgment on what is the most promising realistic outcome.
The decision for me personally is whether to just vote today for Hilary Clinton, or to wait for election day (and keep open the possibility of hell freezing over?).

No comments:

Post a Comment