RSS

AI Ser­vices (CAIS)

TagLast edit: 12 Jul 2020 15:47 UTC by Kaj_Sotala

An AI service as used in the context of Eric Drexler’s technical report Reframing Superintelligence: Comprehensive AI Services as General Intelligence (CAIS), is an AI system that delivers bounded results for some task using bounded resources in bounded time. It is contrasted with agentive AGI, which carries out open-ended goals over an unbounded period of time.

A gradual accumulation of increasingly competent services is one model of how AI might develop. For a summary, see this post.

Refram­ing Su­per­in­tel­li­gence: Com­pre­hen­sive AI Ser­vices as Gen­eral Intelligence

Rohin Shah8 Jan 2019 7:12 UTC
121 points
77 comments5 min readLW link2 reviews
(www.fhi.ox.ac.uk)

Com­ments on CAIS

Richard_Ngo12 Jan 2019 15:20 UTC
76 points
14 comments7 min readLW link

Drexler on AI Risk

PeterMcCluskey1 Feb 2019 5:11 UTC
35 points
10 comments9 min readLW link
(www.bayesianinvestor.com)

Robin Han­son on Lump­iness of AI Services

DanielFilan17 Feb 2019 23:08 UTC
15 points
2 comments2 min readLW link
(www.overcomingbias.com)

[Question] What are CAIS’ bold­est near/​medium-term pre­dic­tions?

jacobjacob28 Mar 2019 13:14 UTC
31 points
17 comments1 min readLW link

Six AI Risk/​Strat­egy Ideas

Wei Dai27 Aug 2019 0:40 UTC
64 points
17 comments4 min readLW link1 review

[Link] Book Re­view: Refram­ing Su­per­in­tel­li­gence (SSC)

ioannes28 Aug 2019 22:57 UTC
46 points
9 comments2 min readLW link

AI Ser­vices as a Re­search Paradigm

VojtaKovarik20 Apr 2020 13:00 UTC
30 points
12 comments4 min readLW link
(docs.google.com)

Box in­ver­sion hypothesis

Jan Kulveit20 Oct 2020 16:20 UTC
59 points
4 comments3 min readLW link

Del­e­gated agents in prac­tice: How com­pa­nies might end up sel­l­ing AI ser­vices that act on be­half of con­sumers and coal­i­tions, and what this im­plies for safety research

Remmelt26 Nov 2020 11:17 UTC
7 points
3 comments4 min readLW link

They gave LLMs ac­cess to physics simulators

ryan_b17 Oct 2022 21:21 UTC
50 points
18 comments1 min readLW link
(arxiv.org)

The econ­omy as an anal­ogy for ad­vanced AI systems

15 Nov 2022 11:16 UTC
28 points
0 comments5 min readLW link

Take 6: CAIS is ac­tu­ally Or­wellian.

Charlie Steiner7 Dec 2022 13:50 UTC
14 points
8 comments2 min readLW link

CAIS-in­spired ap­proach to­wards safer and more in­ter­pretable AGIs

Peter Hroššo27 Mar 2023 14:36 UTC
13 points
7 comments1 min readLW link

Re­cur­sive Mid­dle Man­ager Hell: AI Edition

VojtaKovarik4 May 2023 20:08 UTC
30 points
11 comments2 min readLW link

Up­dat­ing Drexler’s CAIS model

Matthew Barnett16 Jun 2023 22:53 UTC
46 points
32 comments4 min readLW link

“Refram­ing Su­per­in­tel­li­gence” + LLMs + 4 years

Eric Drexler10 Jul 2023 13:42 UTC
116 points
8 comments12 min readLW link

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

particlemania28 Aug 2023 22:19 UTC
4 points
0 comments16 min readLW link

Box in­ver­sion revisited

Jan_Kulveit7 Nov 2023 11:09 UTC
38 points
3 comments8 min readLW link

Open Agency model can solve the AI reg­u­la­tion dilemma

Roman Leventov8 Nov 2023 20:00 UTC
22 points
1 comment2 min readLW link

Elic­it­ing La­tent Knowl­edge in Com­pre­hen­sive AI Ser­vices Models

acabodi17 Nov 2023 2:36 UTC
6 points
0 comments5 min readLW link