(Note date change due to Valentines day)
Thank you for another great meeting in the AISafety.com Reading Group.
For session 83, we will read “Goodhart Taxonomy” by Scott Garrabrant:
https://www.lesserwrong.com/posts/EbFABnst8LsidYs5Y/goodhart-taxonomy
Join us by Skype (add “soeren.elverlin”) on Tuesday the 13th of February 19:45 UTC.
Thank you for another great meeting in the AISafety.com Reading Group.
For session 82, we will read “An AI Race for Strategic Advantage: Rhetoric and Risks” by Seán S ÓhÉigeartaigh and Stephen Cave:
http://www.aies-conference.com/wp-content/papers/main/AIES_2018_paper_163.pdf
Join us by Skype (add “soeren.elverlin”) on Wednesday the 7th of February 19:45 UTC.
Thank you for another great meeting in the AISafety.com Reading Group.
For session 81, we will read “Intelligence Explosion?” and “Reply to Bostrom’s arguments for a hard takeoff” by Brian Thomasik:
https://foundational-research.org/artificial-intelligence-and-its-implications-for-future-suffering#Intelligence_explosion
Join us by Skype (add “soeren.elverlin”) on Wednesday the 31st of January 19:45 UTC.
Thank you for another great meeting in the AISafety.com Reading Group.
For session 80, we will read “Superintelligence As a Cause or Cure For Risks of Astronomical Suffering” by Kaj Sotala and Lukas Gloor.
http://www.informatica.si/index.php/informatica/article/view/1877/1098
Join us by Skype (add “soeren.elverlin”) on Wednesday the 24th of January 19:45 UTC.
Thank you for another great meeting in the AISafety.com Reading Group.
For session 79, we will read “Impossibility of deducing preferences and rationality from human policy” by Stuart Armstrong and Sören Mindermann:
https://arxiv.org/pdf/1712.05812.pdf
Stuart Armstrong will be present, and give a presentation of his work.
Join us by Skype (add “soeren.elverlin”) on Wednesday the 17th of January 19:45 UTC.
Thank you for another great meeting in the AISafety.com Reading Group.
For session 78, we will read “On the Promotion of Safe and Socially Beneficial Artificial Intelligence” by Seth D. Baum:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2816323
Join us by Skype (add “soeren.elverlin”) on Tuesday the 9th of January 19:45 UTC.
I have made a map of the AI Safety Community.
More On Dota 2 by Open Ai – Timeline of the DOTA-bot’s rapid improvement. Bot Exploits. Physical Infrastructure. What needs to be done to play 5×5.
Openai Bots Were Defeated At Least 50 Times – People could play against the openAI Dota bot. Several people found strategies to beat the bot. One of the human victors explains their strategy.
Dota 2 by Open Ai – Open AI codes a 1v1 Dota-2 bot that defeaated top players. The bots actions per minute were comparable to many humans. The bot learned the game from scratch by self-play, and does not use imitation learning or tree search. The game involves hidden information and the bot’s strategies were complicated.