(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)
Re52: Big Questions About AI
Retraice^1
WAAIT Part 3: What individuals, humanity, and players in the game can ask themselves.
Air date: Wednesday, 16th Nov. 2022, 11:00 PM Eastern/US.
AlphaCode
This new AI system codes better than 50% of human coders in competitions.^2 How should we think about big advances in AI?
o amplification of something: a kind of tool, now more powerful than before; o arrival of something: a strange new colleague, or an alien.^3
Individuals:
How much do I matter?
One way of thinking about an individual's (or group's) power is the Banzhaf power index:^4 Of all the possible voting scenarios, in how many does this individual have the capacity to swing the outcome one way or another?
The harsh truth about AI, as regards the future of humanity, is:
* The incapable will remain irrelevant.
* The ignorant will be ignored.
But the group of people who are powerful, with respect to AI, is not a fixed entity: new arrivals are always joining; old members are always leaving.
Individuals who do not want to be irrelevant or ignored will have to become capable in AI or very well-informed about it. Those who will become capable, and/or well-informed, will be the most motivated to do so.
Humanity:
What are we doing again? Where are we headed?
As a whole, humanity can ask the question: AI is for what?^5 Some answers include:
* specific tasks;
* specific knowledge;
* specific actions;
* actions that go beyond satiable needs and wants, to `wants' like everything-better-more-all-the-time.
Players:
How to win?
How can humans possibly respond to the forces of nature creating unfriendly AI--either autonomous colleague-AI or tool-AI wielded by the bad guys? Perhaps by trying to prevent it, or by trying to understand it in advance well enough to manage the risks?
Setting aside the possibility of colleague-AI, tool-AI is going to keep getting stronger and stronger. Are you a cosmist, who believes we should keep building it no matter what? Or a terran, who refuses to build things that pose a catastrophic risk to humanity?^6
If there is a general problem with our ability to respond to these possibilities, is it based on:
* the technical threat of AI?
* our capacity to respond itself?
+ If it is our capacity, is the problem IQ or trust? Or something else?
And is AI the most important force in nature to fear? What about what we've already seen from humans?
And once the key problems are identified: What would have to change, to fix the response problem?
_
References
Brockman, J. (Ed.) (2019). Possible Minds: Twenty-Five Ways of Looking at AI. Penguin. ISBN: 978-0525557999. Searches:
https://www.amazon.com/s?k=978-0525557999
https://www.google.com/search?q=isbn+978-0525557999
https://lccn.loc.gov/2018032888
de Garis, H. (2005). The Artilect War: Cosmists vs. Terrans: A Bitter Controversy Concerning Whether Humanity Should Build Godlike Massively Intelligent Machines. ETC Publications. ISBN: 0882801546. Searches:
https://www.amazon.com/s?k=0882801546
https://www.google.com/search?q=isbn+0882801546
Dennett, D. C. (2019). What can we do? (pp. 41-53). In Brockman (2019).
Retraice (2022/11/07). Re43: The Midterms -- Part 1. retraice.com.
https://www.retraice.com/segments/re43 Retrieved 9th Nov. 2022.
Russell, S. (2019). Human Compatible: Artificial Intelligence and the Problem of Control. Viking. ISBN: 978-0525558613. Searches:
https://www.amazon.com/s?k=978-0525558613
https://www.google.com/search?q=isbn+978-0525558613
https://lccn.loc.gov/2019029688
Footnotes
^1 https://www.retraice.com/retraice
^2 Google's DeepMind says its AI coding bot is `competitive' with humans, theregister.com, Simon Sharwood, Feb 3rd, 2022.
^3 Dennett (2019) says we should stick to building tools, not colleagues. Russell (2019) likens super AI to the impending arrival of an alien species (p. 3).
^4 Retraice (2022/11/07).
^5 For some reason, this phrasing is better out-loud than `What is AI for?'
^6 de Garis (2005).
