RSS

Com­pre­hen­sive AI Services

TagLast edit: 8 Jan 2022 15:32 UTC by Pablo

Comprehensive AI services (CAIS) is an approach to understanding superintelligence proposed by Eric Drexler. In Drexler’s words, “CAIS provides a model of flexible, general intelligence in which agents are a class of service-providing products, rather than a natural or necessary engine of progress in themselves.”[1] On this approach, advanced AI systems are conceived as “services”, or systems with an expanding set of capabilities, one of which is the optional capability of agency.

Further reading

Alexander, Scott (2019) Review of Eric Drexler, Reframing Superintelligence: Comprehensive AI Services as General Intelligence, Slate Star Codex, August 27.

Baumann, Tobias (2020) Summary of Eric Drexler’s work on reframing AI safety, Reducing Risks of Future Suffering, May 21.
A clear bullet-point summary of Drexler 2019.

Drexler, K. Eric (2019) Reframing superintelligence: Comprehensive AI services as general intelligence, technical report #2019-1, Future of Humanity Institute, University of Oxford.

Hanson, Robin (2019) How lumpy AI services?, Overcoming Bias, February 14.

McCluskey, Peter (2019) Drexler on AI risk, Bayesian Investor Blog, January 30.

Ngo, Richard (2019) Comments on CAIS, LessWrong, January 12.

Shah, Rohin (2019) Reframing superintelligence: Comprehensive AI services as general intelligence, LessWrong, January 8.
A very clear summary of the CAIS approach.

Related entries

AI alignment | Eric Drexler | superintelligence | tool AI

  1. ^

    Drexler, K. Eric (2019) Reframing superintelligence: Comprehensive AI services as general intelligence, technical report #2019-1, Future of Humanity Institute, University of Oxford.

Eric Drexler: Refram­ing Superintelligence

EA Global1 Apr 2019 15:49 UTC
16 points
0 comments18 min readEA link
(www.youtube.com)

De­com­pos­ing Agency — ca­pa­bil­ities with­out desires

Owen Cotton-Barratt11 Jul 2024 9:38 UTC
37 points
2 comments12 min readEA link
(strangecities.substack.com)

[Question] Will more AI sys­tems be trained to make use of pre­ex­ist­ing com­pu­ta­tional tools?

deep6 May 2022 17:16 UTC
5 points
4 comments1 min readEA link

Open Agency model can solve the AI reg­u­la­tion dilemma

Roman Leventov9 Nov 2023 15:22 UTC
4 points
0 comments2 min readEA link

Paradigms and The­ory Choice in AI: Adap­tivity, Econ­omy and Control

particlemania28 Aug 2023 22:44 UTC
3 points
0 comments16 min readEA link

AGI will be made of het­ero­ge­neous com­po­nents, Trans­former and Selec­tive SSM blocks will be among them

Roman Leventov27 Dec 2023 14:51 UTC
5 points
0 comments1 min readEA link
No comments.