RSS

AI forecasting

TagLast edit: 22 Jul 2022 21:02 UTC by Leo

AI forecasting is the process of trying to predict a variety of outcomes related to AI, such as when human-level AI will emerge (AI timelines) and what impacts it will have.

Further reading

Cotra, Ajeya (2020) Draft report on AI timelines, LessWrong, September 18.

Davidson, Tom (2021) Semi-informative priors over AI timelines, Open Philanthropy, March 25.

Gruetzemacher, Ross et al. (2020) Forecasting AI progress: a research agenda, arXiv:2008.01848 [Cs].

Koehler, Arden, Robert Wiblin & Keiran Harris (2020) Danny Hernandez on forecasting and the drivers of AI progress, 80,000 Hours, May 22.

Wiblin, Robert & Keiran Harris (2018) How well can we actually predict the future? Katja Grace on why expert opinion isn’t a great guide to AI’s impact and how to do better, 80,000 Hours, August 21.

External links

AI Forecasting Dictionary. A set of standards and conventions for precisely interpreting AI and auxiliary terms.

AI timelines. List of publications by AI Impacts.

Related entries

AI governance | AI alignment | AI Impacts | estimation of existential risk | forecasting | long-range forecasting | quantum computing | tabletop exercises

AI timelines by bio an­chors: the de­bate in one place

Will Aldred30 Jul 2022 23:04 UTC
93 points
6 comments2 min readEA link

AI Fore­cast­ing Re­search Ideas

Jaime Sevilla17 Nov 2022 17:37 UTC
78 points
1 comment1 min readEA link
(docs.google.com)

AI Timelines: Where the Ar­gu­ments, and the “Ex­perts,” Stand

Holden Karnofsky7 Sep 2021 17:35 UTC
88 points
3 comments11 min readEA link

Liter­a­ture re­view of Trans­for­ma­tive Ar­tifi­cial In­tel­li­gence timelines

Jaime Sevilla27 Jan 2023 20:36 UTC
148 points
10 comments1 min readEA link

AGI and the EMH: mar­kets are not ex­pect­ing al­igned or un­al­igned AI in the next 30 years

basil.halperin10 Jan 2023 16:05 UTC
333 points
175 comments26 min readEA link

[Our World in Data] AI timelines: What do ex­perts in ar­tifi­cial in­tel­li­gence ex­pect for the fu­ture? (Roser, 2023)

Will Aldred7 Feb 2023 14:52 UTC
89 points
1 comment1 min readEA link
(ourworldindata.org)

A con­cern about the “evolu­tion­ary an­chor” of Ajeya Co­tra’s re­port on AI timelines.

NunoSempere16 Aug 2022 14:44 UTC
75 points
40 comments5 min readEA link
(nunosempere.com)

Samotsvety’s AI risk forecasts

elifland9 Sep 2022 4:01 UTC
175 points
30 comments3 min readEA link

AI Risk & Policy Fore­casts from Me­tac­u­lus & FLI’s AI Path­ways Workshop

Will Aldred16 May 2023 8:53 UTC
41 points
0 comments8 min readEA link

Re­port on Whether AI Could Drive Ex­plo­sive Eco­nomic Growth

Tom_Davidson25 Jun 2021 23:02 UTC
63 points
5 comments2 min readEA link

A com­pute-based frame­work for think­ing about the fu­ture of AI

Matthew_Barnett31 May 2023 22:00 UTC
96 points
36 comments19 min readEA link

[Link post] How plau­si­ble are AI Takeover sce­nar­ios?

SammyDMartin27 Sep 2021 13:03 UTC
26 points
0 comments1 min readEA link

AI X-Risk: In­te­grat­ing on the Shoulders of Giants

TD_Pilditch1 Nov 2022 16:07 UTC
34 points
0 comments47 min readEA link

Two con­trast­ing mod­els of “in­tel­li­gence” and fu­ture growth

Magnus Vinding24 Nov 2022 11:54 UTC
74 points
32 comments22 min readEA link

Oper­a­tional­iz­ing timelines

Zach Stein-Perlman10 Mar 2023 17:30 UTC
30 points
2 comments1 min readEA link

Sur­vey on AI ex­is­ten­tial risk scenarios

Sam Clarke8 Jun 2021 17:12 UTC
154 points
11 comments6 min readEA link

Will scal­ing work?

Vasco Grilo4 Feb 2024 9:29 UTC
19 points
1 comment12 min readEA link
(www.dwarkeshpatel.com)

Me­tac­u­lus is build­ing a team ded­i­cated to AI forecasting

christian18 Oct 2022 16:08 UTC
35 points
0 comments1 min readEA link
(apply.workable.com)

Pre­dictable up­dat­ing about AI risk

Joe_Carlsmith8 May 2023 22:05 UTC
130 points
12 comments36 min readEA link

What a com­pute-cen­tric frame­work says about AI take­off speeds

Tom_Davidson23 Jan 2023 4:09 UTC
189 points
7 comments16 min readEA link
(www.lesswrong.com)

Why AI is Harder Than We Think—Me­lanie Mitchell

BrownHairedEevee28 Apr 2021 8:19 UTC
45 points
7 comments2 min readEA link
(arxiv.org)

Biolog­i­cal An­chors ex­ter­nal re­view by Jen­nifer Lin (linkpost)

peterhartree30 Nov 2022 13:06 UTC
36 points
0 comments1 min readEA link
(docs.google.com)

Epoch AI is Hiring an Eco­nomics of AI Researcher

merilalama3 May 2024 0:03 UTC
24 points
2 comments3 min readEA link
(careers.rethinkpriorities.org)

Will AI end ev­ery­thing? A guide to guess­ing | EAG Bay Area 23

Katja_Grace25 May 2023 17:01 UTC
74 points
1 comment21 min readEA link

[Question] Ask­ing for on­line re­sources why AI now is near AGI

jackchang11018 May 2023 0:04 UTC
6 points
4 comments1 min readEA link

Prac­tic­ing my Hand­writ­ing in 1439

Maxwell Tabarrok3 Feb 2024 13:22 UTC
19 points
1 comment3 min readEA link
(www.maximum-progress.com)

AGI Timelines in Gover­nance: Differ­ent Strate­gies for Differ­ent Timeframes

simeon_c19 Dec 2022 21:31 UTC
110 points
19 comments1 min readEA link

The repli­ca­tion and em­u­la­tion of GPT-3

Ben Cottier21 Dec 2022 13:49 UTC
14 points
0 comments33 min readEA link

When re­port­ing AI timelines, be clear who you’re defer­ring to

Sam Clarke10 Oct 2022 14:24 UTC
120 points
20 comments1 min readEA link

Trans­for­ma­tive AGI by 2043 is <1% likely

Ted Sanders6 Jun 2023 15:51 UTC
92 points
92 comments5 min readEA link
(arxiv.org)

[Question] Why most peo­ple in EA are con­fi­dent that AI will sur­pass hu­mans?

jackchang11025 May 2023 13:39 UTC
2 points
6 comments1 min readEA link

Ex­plor­ing Me­tac­u­lus’s AI Track Record

Peter Scoblic1 May 2023 21:02 UTC
52 points
5 comments7 min readEA link
(www.metaculus.com)

Without a tra­jec­tory change, the de­vel­op­ment of AGI is likely to go badly

Max H30 May 2023 0:21 UTC
1 point
0 comments13 min readEA link

Fore­cast in the Un­der­stand­ing AI Series With Ti­mothy B. Lee

christian28 Mar 2024 22:27 UTC
12 points
2 comments1 min readEA link
(www.metaculus.com)

Why I think it’s im­por­tant to work on AI forecasting

Matthew_Barnett27 Feb 2023 21:24 UTC
179 points
10 comments10 min readEA link

[linkpost] “What Are Rea­son­able AI Fears?” by Robin Han­son, 2023-04-23

Arjun Panickssery14 Apr 2023 23:26 UTC
41 points
3 comments4 min readEA link
(quillette.com)

Epoch is hiring a Product and Data Vi­su­al­iza­tion Designer

merilalama25 Nov 2023 0:14 UTC
21 points
0 comments4 min readEA link
(careers.rethinkpriorities.org)

In­trin­sic limi­ta­tions of GPT-4 and other large lan­guage mod­els, and why I’m not (very) wor­ried about GPT-n

Fods123 Jun 2023 13:09 UTC
28 points
3 comments11 min readEA link

Ques­tions for fur­ther in­ves­ti­ga­tion of AI diffusion

Ben Cottier21 Dec 2022 13:50 UTC
28 points
0 comments11 min readEA link

AI Policy In­sights from the AIMS Survey

Janet Pauketat22 Feb 2024 19:17 UTC
10 points
1 comment18 min readEA link
(www.sentienceinstitute.org)

Drivers of large lan­guage model diffu­sion: in­cre­men­tal re­search, pub­lic­ity, and cascades

Ben Cottier21 Dec 2022 13:50 UTC
21 points
0 comments29 min readEA link

Scor­ing fore­casts from the 2016 “Ex­pert Sur­vey on Progress in AI”

PatrickL1 Mar 2023 14:39 UTC
204 points
21 comments9 min readEA link

How Do AI Timelines Affect Giv­ing Now vs. Later?

MichaelDickens3 Aug 2021 3:36 UTC
36 points
8 comments8 min readEA link

[Question] What con­sid­er­a­tions in­fluence whether I have more in­fluence over short or long timelines?

kokotajlod5 Nov 2020 19:57 UTC
18 points
0 comments1 min readEA link

Effects of anti-ag­ing re­search on the long-term future

Matthew_Barnett27 Feb 2020 22:42 UTC
61 points
33 comments4 min readEA link

Fore­cast­ing Trans­for­ma­tive AI: Are we “trend­ing to­ward” trans­for­ma­tive AI? (How would we know?)

Holden Karnofsky24 Aug 2021 17:15 UTC
56 points
12 comments10 min readEA link

“Biolog­i­cal an­chors” is about bound­ing, not pin­point­ing, AI timelines

Holden Karnofsky18 Nov 2021 21:03 UTC
38 points
9 comments11 min readEA link

It takes 5 lay­ers and 1000 ar­tifi­cial neu­rons to simu­late a sin­gle biolog­i­cal neu­ron [Link]

MichaelStJules7 Sep 2021 21:53 UTC
44 points
17 comments2 min readEA link

Pre­dict re­sponses to the “ex­is­ten­tial risk from AI” survey

RobBensinger28 May 2021 1:38 UTC
36 points
8 comments2 min readEA link

Disagree­ment with bio an­chors that lead to shorter timelines

mariushobbhahn16 Nov 2022 14:40 UTC
85 points
1 comment1 min readEA link

[Question] Is this a good way to bet on short timelines?

kokotajlod28 Nov 2020 14:31 UTC
17 points
16 comments1 min readEA link

Is AI fore­cast­ing a waste of effort on the mar­gin?

Emrik5 Nov 2022 0:41 UTC
10 points
6 comments3 min readEA link

Dis­cussing how to al­ign Trans­for­ma­tive AI if it’s de­vel­oped very soon

elifland28 Nov 2022 16:17 UTC
36 points
0 comments1 min readEA link

My thoughts on nan­otech­nol­ogy strat­egy re­search as an EA cause area

Ben Snodin2 May 2022 9:41 UTC
137 points
17 comments33 min readEA link

Publi­ca­tion de­ci­sions for large lan­guage mod­els, and their impacts

Ben Cottier21 Dec 2022 13:50 UTC
14 points
0 comments16 min readEA link

Back­ground for “Un­der­stand­ing the diffu­sion of large lan­guage mod­els”

Ben Cottier21 Dec 2022 13:49 UTC
12 points
0 comments23 min readEA link

Ar­gu­ment Against Im­pact: EU Is Not an AI Su­per­power

EU AI Governance31 Jan 2022 9:48 UTC
35 points
9 comments3 min readEA link

Rele­vant pre-AGI possibilities

kokotajlod20 Jun 2020 13:15 UTC
22 points
0 comments1 min readEA link
(aiimpacts.org)

Birds, Brains, Planes, and AI: Against Ap­peals to the Com­plex­ity/​Mys­te­ri­ous­ness/​Effi­ciency of the Brain

kokotajlod18 Jan 2021 12:39 UTC
27 points
2 comments1 min readEA link

Con­clu­sion and Bibliog­ra­phy for “Un­der­stand­ing the diffu­sion of large lan­guage mod­els”

Ben Cottier21 Dec 2022 13:50 UTC
12 points
0 comments11 min readEA link

[Question] Fore­cast­ing thread: How does AI risk level vary based on timelines?

elifland14 Sep 2022 23:56 UTC
47 points
8 comments1 min readEA link

Win­ners of the EA Crit­i­cism and Red Team­ing Contest

Lizka1 Oct 2022 1:50 UTC
226 points
41 comments19 min readEA link

AI Fore­cast­ing Re­s­olu­tion Coun­cil (Fore­cast­ing in­fras­truc­ture, part 2)

jacobjacob29 Aug 2019 17:43 UTC
28 points
0 comments3 min readEA link

Shul­man and Yud­kowsky on AI progress

CarlShulman4 Dec 2021 11:37 UTC
46 points
0 comments20 min readEA link

Some thoughts on David Rood­man’s model of eco­nomic growth and its re­la­tion to AI timelines

Tom_Davidson19 Jul 2021 21:47 UTC
96 points
6 comments6 min readEA link

Re­port on Semi-in­for­ma­tive Pri­ors for AI timelines (Open Philan­thropy)

Tom_Davidson26 Mar 2021 17:46 UTC
62 points
6 comments2 min readEA link

Im­pli­ca­tions of large lan­guage model diffu­sion for AI governance

Ben Cottier21 Dec 2022 13:50 UTC
14 points
0 comments38 min readEA link

Re­place­ment for PONR concept

kokotajlod2 Sep 2022 0:38 UTC
14 points
1 comment3 min readEA link

Con­ver­sa­tion on tech­nol­ogy fore­cast­ing and gradualism

RobBensinger9 Dec 2021 19:00 UTC
15 points
3 comments31 min readEA link

[Question] What is the best source to ex­plain short AI timelines to a skep­ti­cal per­son?

trevor123 Nov 2022 5:20 UTC
2 points
3 comments1 min readEA link

Ajeya’s TAI timeline short­ened from 2050 to 2040

Zach Stein-Perlman3 Aug 2022 0:00 UTC
59 points
2 comments1 min readEA link
(www.lesswrong.com)

Vignettes Work­shop (AI Im­pacts)

kokotajlod15 Jun 2021 11:02 UTC
43 points
5 comments1 min readEA link

A Bird’s Eye View of the ML Field [Prag­matic AI Safety #2]

ThomasW9 May 2022 17:15 UTC
97 points
2 comments36 min readEA link

[Question] What are the top pri­ori­ties in a slow-take­off, mul­ti­po­lar world?

JP Addison25 Aug 2021 8:47 UTC
26 points
9 comments1 min readEA link

AGI x-risk timelines: 10% chance (by year X) es­ti­mates should be the head­line, not 50%.

Greg_Colbourn1 Mar 2022 12:02 UTC
67 points
22 comments1 min readEA link

AI Fore­cast­ing Dic­tionary (Fore­cast­ing in­fras­truc­ture, part 1)

jacobjacob8 Aug 2019 13:16 UTC
18 points
0 comments5 min readEA link

“Slower tech de­vel­op­ment” can be about or­der­ing, grad­u­al­ness, or dis­tance from now

MichaelA14 Nov 2021 20:58 UTC
47 points
3 comments4 min readEA link

Against GDP as a met­ric for timelines and take­off speeds

kokotajlod29 Dec 2020 17:50 UTC
47 points
6 comments14 min readEA link

[Question] How can I bet on short timelines?

kokotajlod7 Nov 2020 12:45 UTC
33 points
12 comments2 min readEA link

Fore­cast­ing Trans­for­ma­tive AI: What Kind of AI?

Holden Karnofsky10 Aug 2021 21:38 UTC
62 points
3 comments10 min readEA link

AI Fore­cast­ing Ques­tion Database (Fore­cast­ing in­fras­truc­ture, part 3)

jacobjacob3 Sep 2019 14:57 UTC
23 points
2 comments5 min readEA link

Draft re­port on ex­is­ten­tial risk from power-seek­ing AI

Joe_Carlsmith28 Apr 2021 21:41 UTC
87 points
34 comments1 min readEA link

Atari early

AI Impacts2 Apr 2020 23:28 UTC
34 points
2 comments1 min readEA link
(aiimpacts.org)

My take on What We Owe the Future

elifland1 Sep 2022 18:07 UTC
353 points
51 comments26 min readEA link

On Defer­ence and Yud­kowsky’s AI Risk Estimates

bgarfinkel19 Jun 2022 14:35 UTC
285 points
194 comments17 min readEA link

Biol­ogy-In­spired AGI Timelines: The Trick That Never Works

EliezerYudkowsky1 Dec 2021 22:44 UTC
22 points
3 comments66 min readEA link

My at­tempt to think about AI timelines

Ben Snodin18 May 2021 17:05 UTC
53 points
20 comments8 min readEA link

Dis­con­tin­u­ous progress in his­tory: an update

AI Impacts17 Apr 2020 16:28 UTC
69 points
3 comments24 min readEA link

Phil Tram­mell on Eco­nomic Growth Un­der Trans­for­ma­tive AI

Michaël Trazzi24 Oct 2021 18:10 UTC
10 points
0 comments54 min readEA link
(youtu.be)

Per­sua­sion Tools: AI takeover with­out AGI or agency?

kokotajlod20 Nov 2020 16:56 UTC
15 points
5 comments10 min readEA link

AMA: Ajeya Co­tra, re­searcher at Open Phil

Ajeya28 Jan 2021 17:38 UTC
84 points
105 comments1 min readEA link

Asya Ber­gal: Rea­sons you might think hu­man-level AI is un­likely to hap­pen soon

EA Global26 Aug 2020 16:01 UTC
24 points
2 comments17 min readEA link
(www.youtube.com)

Draft re­port on AI timelines

Ajeya15 Dec 2020 12:10 UTC
35 points
0 comments1 min readEA link
(alignmentforum.org)

Fore­cast­ing trans­for­ma­tive AI: the “biolog­i­cal an­chors” method in a nutshell

Holden Karnofsky31 Aug 2021 18:17 UTC
50 points
13 comments18 min readEA link

Im­pli­ca­tion of AI timelines on plan­ning and solutions

JJ Hepburn21 Aug 2021 5:11 UTC
15 points
0 comments2 min readEA link

Fore­cast­ing trans­for­ma­tive AI: what’s the bur­den of proof?

Holden Karnofsky17 Aug 2021 17:14 UTC
71 points
17 comments16 min readEA link

An­nounc­ing Epoch: A re­search or­ga­ni­za­tion in­ves­ti­gat­ing the road to Trans­for­ma­tive AI

Jaime Sevilla27 Jun 2022 13:39 UTC
183 points
11 comments2 min readEA link
(epochai.org)

On AI and Compute

johncrox3 Apr 2019 21:26 UTC
39 points
12 comments5 min readEA link

What 2026 looks like (Daniel’s me­dian fu­ture)

kokotajlod7 Aug 2021 5:14 UTC
38 points
1 comment2 min readEA link
(www.lesswrong.com)

Com­par­ing top fore­cast­ers and do­main experts

Gavin6 Mar 2022 20:43 UTC
205 points
40 comments3 min readEA link

More Chris­ti­ano, Co­tra, and Yud­kowsky on AI progress

EliezerYudkowsky6 Dec 2021 20:34 UTC
16 points
1 comment40 min readEA link

“Ex­is­ten­tial risk from AI” sur­vey results

RobBensinger1 Jun 2021 20:19 UTC
80 points
35 comments11 min readEA link

‘Dis­solv­ing’ AI Risk – Pa­ram­e­ter Uncer­tainty in AI Fu­ture Forecasting

Froolow18 Oct 2022 22:54 UTC
109 points
63 comments39 min readEA link

Have your timelines changed as a re­sult of ChatGPT?

Chris Leong5 Dec 2022 15:03 UTC
30 points
18 comments1 min readEA link

AI timelines and the­o­ret­i­cal un­der­stand­ing of deep learn­ing

Venky102412 Sep 2021 16:26 UTC
4 points
8 comments2 min readEA link

AGI Isn’t Close—Fu­ture Fund Wor­ld­view Prize

Toni MUENDEL18 Dec 2022 16:03 UTC
−8 points
24 comments13 min readEA link

[Question] Ques­tions on databases of AI Risk estimates

Froolow2 Oct 2022 9:12 UTC
24 points
12 comments2 min readEA link

Long-Term Fu­ture Fund: May 2021 grant recommendations

abergal27 May 2021 6:44 UTC
110 points
17 comments58 min readEA link

[Link post] Pa­ram­e­ter counts in Ma­chine Learning

Jaime Sevilla1 Jul 2021 15:44 UTC
15 points
0 comments1 min readEA link

Un­der­stand­ing the diffu­sion of large lan­guage mod­els: summary

Ben Cottier21 Dec 2022 13:49 UTC
127 points
18 comments22 min readEA link

Chris­ti­ano, Co­tra, and Yud­kowsky on AI progress

Ajeya25 Nov 2021 16:30 UTC
18 points
6 comments69 min readEA link

[Question] What will be some of the most im­pact­ful ap­pli­ca­tions of ad­vanced AI in the near term?

IanDavidMoss3 Mar 2022 15:26 UTC
16 points
7 comments1 min readEA link

Sum­mary of Si­tu­a­tional Aware­ness—The Decade Ahead

OscarD8 Jun 2024 11:29 UTC
137 points
5 comments18 min readEA link

[Link] “The AI Timelines Scam”

Milan_Griffes11 Jul 2019 3:37 UTC
22 points
2 comments1 min readEA link

[Question] What are the num­bers in mind for the su­per-short AGI timelines so many long-ter­mists are alarmed about?

Evan_Gaensbauer19 Apr 2022 21:09 UTC
41 points
2 comments1 min readEA link

Does gen­er­al­ity pay? GPT-3 can provide pre­limi­nary ev­i­dence.

BrownHairedEevee12 Jul 2020 18:53 UTC
21 points
4 comments2 min readEA link

Yud­kowsky and Chris­ti­ano dis­cuss “Take­off Speeds”

EliezerYudkowsky22 Nov 2021 19:42 UTC
42 points
0 comments60 min readEA link

When you plan ac­cord­ing to your AI timelines, should you put more weight on the me­dian fu­ture, or the me­dian fu­ture | even­tual AI al­ign­ment suc­cess? ⚖️

Jeffrey Ladish5 Jan 2023 1:55 UTC
16 points
2 comments2 min readEA link

Up­date to Samotsvety AGI timelines

Misha_Yagudin24 Jan 2023 4:27 UTC
120 points
9 comments4 min readEA link

Un­jour­nal: Eval­u­a­tions of “Ar­tifi­cial In­tel­li­gence and Eco­nomic Growth”, and new host­ing space

david_reinstein17 Mar 2023 20:20 UTC
47 points
0 comments2 min readEA link
(unjournal.pubpub.org)

Trends in the dol­lar train­ing cost of ma­chine learn­ing systems

Ben Cottier1 Feb 2023 14:48 UTC
63 points
3 comments1 min readEA link

Tech­nolog­i­cal de­vel­op­ments that could in­crease risks from nu­clear weapons: A shal­low review

MichaelA9 Feb 2023 15:41 UTC
79 points
3 comments5 min readEA link
(bit.ly)

Sur­vey on in­ter­me­di­ate goals in AI governance

MichaelA17 Mar 2023 12:44 UTC
155 points
4 comments1 min readEA link

An­nounc­ing the Open Philan­thropy AI Wor­ld­views Contest

Jason Schukraft10 Mar 2023 2:33 UTC
137 points
33 comments3 min readEA link
(www.openphilanthropy.org)

[Linkpost] Scott Alexan­der re­acts to OpenAI’s lat­est post

Akash11 Mar 2023 22:24 UTC
105 points
4 comments1 min readEA link

How bad a fu­ture do ML re­searchers ex­pect?

Katja_Grace13 Mar 2023 5:47 UTC
165 points
20 comments1 min readEA link

Su­perfore­cast­ing the premises in “Is power-seek­ing AI an ex­is­ten­tial risk?”

Joe_Carlsmith18 Oct 2023 20:33 UTC
114 points
3 comments1 min readEA link

Fu­ture Mat­ters #8: Bing Chat, AI labs on safety, and paus­ing Fu­ture Matters

Pablo21 Mar 2023 14:50 UTC
81 points
5 comments24 min readEA link

An­nounc­ing Epoch’s newly ex­panded Pa­ram­e­ters, Com­pute and Data Trends in Ma­chine Learn­ing database

Robi Rahman25 Oct 2023 3:03 UTC
38 points
1 comment1 min readEA link
(epochai.org)

Me­tac­u­lus Pre­dicts Weak AGI in 2 Years and AGI in 10

Chris Leong24 Mar 2023 19:43 UTC
27 points
12 comments1 min readEA link

Clar­ify­ing and pre­dict­ing AGI

richard_ngo4 May 2023 15:56 UTC
69 points
2 comments1 min readEA link

AI Views Snapshots

RobBensinger13 Dec 2023 0:45 UTC
25 points
0 comments1 min readEA link

Con­tin­u­ous doesn’t mean slow

Tom_Davidson10 May 2023 12:17 UTC
64 points
1 comment4 min readEA link

Long list of AI ques­tions

NunoSempere6 Dec 2023 11:12 UTC
124 points
13 comments86 min readEA link

Fo­cus­ing your im­pact on short vs long TAI timelines

kuhanj30 Sep 2023 19:23 UTC
39 points
0 comments10 min readEA link

FLI pod­cast se­ries, “Imag­ine A World”, about as­pira­tional fu­tures with AGI

Jackson Wagner13 Oct 2023 16:03 UTC
18 points
0 comments4 min readEA link

Thoughts on “The Offense-Defense Balance Rarely Changes”

Cullen12 Feb 2024 3:26 UTC
42 points
4 comments5 min readEA link

Me­tac­u­lus Launches Fu­ture of AI Series, Based on Re­search Ques­tions by Arb Research

christian13 Mar 2024 21:14 UTC
29 points
0 comments1 min readEA link
(www.metaculus.com)

Timelines to Trans­for­ma­tive AI: an investigation

Zershaaneh Qureshi25 Mar 2024 18:11 UTC
71 points
8 comments50 min readEA link

On the fu­ture of lan­guage models

Owen Cotton-Barratt20 Dec 2023 16:58 UTC
115 points
3 comments36 min readEA link

Non-al­ign­ment pro­ject ideas for mak­ing trans­for­ma­tive AI go well

Lukas Finnveden4 Jan 2024 7:23 UTC
64 points
1 comment3 min readEA link
(lukasfinnveden.substack.com)

Sur­vey of 2,778 AI au­thors: six parts in pictures

Katja_Grace6 Jan 2024 4:43 UTC
176 points
10 comments1 min readEA link

Lan­guage mod­els sur­prised us

Ajeya29 Aug 2023 21:18 UTC
59 points
10 comments5 min readEA link

Re­sults from an Ad­ver­sar­ial Col­lab­o­ra­tion on AI Risk (FRI)

Forecasting Research Institute11 Mar 2024 15:54 UTC
193 points
25 comments9 min readEA link
(forecastingresearch.org)

An­nounc­ing Epoch’s dash­board of key trends and figures in Ma­chine Learning

Jaime Sevilla13 Apr 2023 7:33 UTC
127 points
4 comments1 min readEA link

AI Im­pacts: His­toric trends in tech­nolog­i­cal progress

Aaron Gertler12 Feb 2020 0:08 UTC
55 points
5 comments3 min readEA link

Take­off speeds pre­sen­ta­tion at Anthropic

Tom_Davidson4 Jun 2024 22:46 UTC
29 points
1 comment1 min readEA link

I made a news site based on pre­dic­tion markets

vandemonian5 Jun 2023 18:33 UTC
226 points
14 comments4 min readEA link

How im­por­tant are ac­cu­rate AI timelines for the op­ti­mal spend­ing sched­ule on AI risk in­ter­ven­tions?

Tristan Cook16 Dec 2022 16:05 UTC
30 points
0 comments6 min readEA link

Pod­cast: Bryan Ca­plan on open bor­ders, UBI, to­tal­i­tar­i­anism, AI, pan­demics, util­i­tar­i­anism and la­bor economics

Gus Docker22 Feb 2022 15:04 UTC
22 points
0 comments46 min readEA link
(www.utilitarianpodcast.com)

“Tech­nolog­i­cal un­em­ploy­ment” AI vs. “most im­por­tant cen­tury” AI: how far apart?

Holden Karnofsky11 Oct 2022 4:50 UTC
17 points
1 comment3 min readEA link
(www.cold-takes.com)

Fore­cast­ing Through Fiction

Yitz6 Jul 2022 5:23 UTC
8 points
3 comments6 min readEA link
(www.lesswrong.com)

Against us­ing stock prices to fore­cast AI timelines

basil.halperin10 Jan 2023 16:04 UTC
18 points
5 comments2 min readEA link

[Ru­mour] Microsoft to in­vest $10B in OpenAI, will re­ceive 75% of prof­its un­til they re­coup in­vest­ment: GPT would be in­te­grated with Office

𝕮𝖎𝖓𝖊𝖗𝖆10 Jan 2023 23:43 UTC
25 points
2 comments1 min readEA link

If slow-take­off AGI is some­what likely, don’t give now

Milan_Griffes23 Jan 2019 20:54 UTC
21 points
33 comments2 min readEA link

2024 State of AI Reg­u­la­tory Landscape

Deric Cheng28 May 2024 12:00 UTC
12 points
1 comment2 min readEA link
(www.convergenceanalysis.org)

[Question] What’s the ex­act way you pre­dict prob­a­bil­ity of AI ex­tinc­tion?

jackchang11013 Jun 2023 15:11 UTC
18 points
7 comments1 min readEA link

AI Alter­na­tive Fu­tures: Ex­plo­ra­tory Sce­nario Map­ping for Ar­tifi­cial In­tel­li­gence Risk—Re­quest for Par­ti­ci­pa­tion [Linkpost]

Kiliank9 May 2022 19:53 UTC
17 points
2 comments8 min readEA link

Against the weird­ness heuris­tic

Eleni_A5 Oct 2022 14:13 UTC
5 points
0 comments2 min readEA link

Fore­cast­ing Our World in Data: The Next 100 Years

AlexLeader1 Feb 2023 22:13 UTC
97 points
8 comments66 min readEA link
(www.metaculus.com)

Epoch Im­pact Re­port 2022

Jaime Sevilla2 Feb 2023 13:09 UTC
81 points
1 comment1 min readEA link
(epochai.org)

[Question] Is there any work on cause pri­ori­ti­za­tion that takes into ac­count timelines be­ing wor­ld­view-de­pen­dent?

Chris Leong31 Oct 2023 2:25 UTC
13 points
2 comments1 min readEA link

War in Taiwan and AI Timelines

Jordan_Schneider24 Aug 2022 2:24 UTC
19 points
3 comments9 min readEA link
(www.chinatalk.media)

[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

AGI in sight: our look at the game board

Andrea_Miotti18 Feb 2023 22:17 UTC
25 points
18 comments1 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

Fore­cast­ing Com­pute—Trans­for­ma­tive AI and Com­pute [2/​4]

lennart1 Oct 2021 8:25 UTC
39 points
6 comments19 min readEA link

Call for Cruxes by Rhyme, a Longter­mist His­tory Con­sul­tancy

Lara_TH1 Mar 2023 10:20 UTC
147 points
6 comments3 min readEA link

De­mon­strate and eval­u­ate risks from AI to so­ciety at the AI x Democ­racy re­search hackathon

Esben Kran19 Apr 2024 14:46 UTC
24 points
0 comments6 min readEA link
(www.apartresearch.com)

Why I think strong gen­eral AI is com­ing soon

porby28 Sep 2022 6:55 UTC
14 points
1 comment1 min readEA link

AI Bench­marks Series — Me­tac­u­lus Ques­tions on Eval­u­a­tions of AI Models Against Tech­ni­cal Benchmarks

christian27 Mar 2024 23:05 UTC
10 points
0 comments1 min readEA link
(www.metaculus.com)

Every­thing’s nor­mal un­til it’s not

Eleni_A10 Mar 2023 1:42 UTC
6 points
0 comments3 min readEA link

An­thropic: Core Views on AI Safety: When, Why, What, and How

jonmenaster9 Mar 2023 17:30 UTC
107 points
6 comments22 min readEA link
(www.anthropic.com)

We Ran an AI Timelines Retreat

Lenny McCline17 May 2022 4:40 UTC
46 points
6 comments3 min readEA link

[Question] Is there a pub­lic tracker de­pict­ing at what dates AI has been able to au­to­mate x% of cog­ni­tive tasks (weighted by 2020 eco­nomic value)?

Mitchell Laughlin17 Feb 2024 4:52 UTC
12 points
4 comments1 min readEA link

“AI pre­dic­tions” (Fu­ture Fund AI Wor­ld­view Prize sub­mis­sion)

ketanrama5 Nov 2022 17:51 UTC
3 points
0 comments3 min readEA link
(medium.com)

Epoch is hiring an As­so­ci­ate Data Analyst

merilalama21 Sep 2023 13:25 UTC
9 points
2 comments3 min readEA link
(careers.rethinkpriorities.org)

2023 Open Philan­thropy AI Wor­ld­views Con­test: Odds of Ar­tifi­cial Gen­eral In­tel­li­gence by 2043

srhoades1014 Mar 2023 20:32 UTC
19 points
0 comments46 min readEA link

[Link post] Will we see fast AI Take­off?

SammyDMartin30 Sep 2021 14:03 UTC
18 points
0 comments1 min readEA link

Trans­for­ma­tive AI and Com­pute—Read­ing List

Frederik Berg4 Sep 2023 6:21 UTC
24 points
0 comments1 min readEA link
(docs.google.com)

Im­mor­tal­ity or death by AGI

ImmortalityOrDeathByAGI24 Sep 2023 9:44 UTC
12 points
2 comments4 min readEA link
(www.lesswrong.com)

China-AI fore­cast­ing

Nathan_Barnard25 Feb 2024 16:47 UTC
10 points
2 comments6 min readEA link

My per­sonal cruxes for work­ing on AI safety

Buck13 Feb 2020 7:11 UTC
135 points
35 comments45 min readEA link

An­nounc­ing New Begin­ner-friendly Book on AI Safety and Risk

Darren McKee25 Nov 2023 15:57 UTC
111 points
9 comments1 min readEA link

Public Opinion on AI Safety: AIMS 2023 and 2021 Summary

Janet Pauketat25 Sep 2023 18:09 UTC
19 points
0 comments3 min readEA link
(www.sentienceinstitute.org)

[Question] How much will pre-trans­for­ma­tive AI speed up R&D?

Ben Snodin31 May 2021 20:20 UTC
23 points
0 comments1 min readEA link

Largest AI model in 2 years from $10B

Péter Drótos24 Oct 2023 15:14 UTC
36 points
0 comments7 min readEA link

A sim­ple way of ex­ploit­ing AI’s com­ing eco­nomic im­pact may be highly-impactful

kuira16 Jul 2023 10:30 UTC
5 points
0 comments2 min readEA link
(www.lesswrong.com)

Epoch is hiring an ML Hard­ware Researcher

merilalama20 Jul 2023 19:08 UTC
29 points
0 comments4 min readEA link
(careers.rethinkpriorities.org)

Carl Shul­man on AI takeover mechanisms (& more): Part II of Dwarkesh Pa­tel in­ter­view for The Lu­nar Society

Alejandro Ortega25 Jul 2023 18:31 UTC
28 points
0 comments5 min readEA link
(www.dwarkeshpatel.com)

Re­vis­it­ing the Evolu­tion An­chor in the Biolog­i­cal An­chors Re­port

Janvi18 Mar 2024 3:01 UTC
13 points
1 comment4 min readEA link

As­ter­isk Magaz­ine Is­sue 03: AI

Alejandro Ortega24 Jul 2023 15:53 UTC
34 points
3 comments1 min readEA link
(asteriskmag.com)

Com­pute Re­search Ques­tions and Met­rics—Trans­for­ma­tive AI and Com­pute [4/​4]

lennart28 Nov 2021 22:18 UTC
18 points
2 comments1 min readEA link

Epoch is hiring a Re­search Data Analyst

merilalama22 Nov 2022 17:34 UTC
21 points
0 comments4 min readEA link
(careers.rethinkpriorities.org)

Hacker-AI – Does it already ex­ist?

Erland Wittkotter7 Nov 2022 14:01 UTC
0 points
1 comment1 min readEA link

No “Zero-Shot” Without Ex­po­nen­tial Data: Pre­train­ing Con­cept Fre­quency Deter­mines Mul­ti­modal Model Performance

Nicholas Kruus14 May 2024 23:57 UTC
36 points
2 comments1 min readEA link
(arxiv.org)

Take­aways from the Me­tac­u­lus AI Progress Tournament

Javier Prieto27 Jul 2023 14:37 UTC
85 points
6 comments4 min readEA link

XPT fore­casts on (some) Direct Ap­proach model inputs

Forecasting Research Institute20 Aug 2023 12:39 UTC
37 points
0 comments9 min readEA link

When Will We Spend Enough to Train Trans­for­ma­tive AI

sn28 Mar 2023 0:41 UTC
3 points
0 comments9 min readEA link

[Event] Join Me­tac­u­lus To­mor­row, March 31st, for Fore­cast Fri­day!

christian30 Mar 2023 20:58 UTC
29 points
1 comment1 min readEA link
(www.metaculus.com)

Defer­ence on AI timelines: sur­vey results

Sam Clarke30 Mar 2023 23:03 UTC
68 points
3 comments2 min readEA link

[Question] How to per­suade a non-CS back­ground per­son to be­lieve AGI is 50% pos­si­ble in 2040?

jackchang1101 Apr 2023 15:27 UTC
1 point
7 comments1 min readEA link

[Question] Pre­dic­tions for fu­ture AI gov­er­nance?

jackchang1102 Apr 2023 16:43 UTC
4 points
1 comment1 min readEA link

[Question] Can we eval­u­ate the “tool ver­sus agent” AGI pre­dic­tion?

Ben_West8 Apr 2023 18:35 UTC
63 points
7 comments1 min readEA link

“Origi­nal­ity is noth­ing but ju­di­cious imi­ta­tion”—Voltaire

Damien Lasseur23 Oct 2022 19:00 UTC
1 point
0 comments1 min readEA link

Con­ver­sa­tion on fore­cast­ing with Vaniver and Ozzie Gooen

jacobjacob30 Jul 2019 11:16 UTC
38 points
4 comments32 min readEA link

AGI al­ign­ment re­sults from a se­ries of al­igned ac­tions

hanadulset27 Dec 2021 19:33 UTC
15 points
1 comment6 min readEA link

GPT-3-like mod­els are now much eas­ier to ac­cess and de­ploy than to develop

Ben Cottier21 Dec 2022 13:49 UTC
22 points
3 comments19 min readEA link

[linkpost] When does tech­ni­cal work to re­duce AGI con­flict make a differ­ence?: Introduction

Anthony DiGiovanni16 Sep 2022 14:35 UTC
31 points
0 comments1 min readEA link
(www.lesswrong.com)

Mis­sion-cor­re­lated in­vest­ing: Ex­am­ples of mis­sion hedg­ing and ‘lev­er­ag­ing’

jh11 Mar 2022 9:33 UTC
25 points
1 comment7 min readEA link

AI Takeover Sce­nario with Scaled LLMs

simeon_c16 Apr 2023 23:28 UTC
29 points
1 comment1 min readEA link

No, the EMH does not im­ply that mar­kets have long AGI timelines

Jakob24 Apr 2023 8:27 UTC
83 points
21 comments8 min readEA link

Le Tem­p­is­tiche delle IA: il di­bat­tito e il punto di vista degli “es­perti”

EA Italy17 Jan 2023 23:30 UTC
1 point
0 comments11 min readEA link

Power laws in Speedrun­ning and Ma­chine Learning

Jaime Sevilla24 Apr 2023 10:06 UTC
48 points
0 comments1 min readEA link

A Guide to Fore­cast­ing AI Science Capabilities

Eleni_A29 Apr 2023 6:51 UTC
19 points
1 comment4 min readEA link

4 Key As­sump­tions in AI Safety

Prometheus7 Nov 2022 10:50 UTC
5 points
0 comments1 min readEA link

[Question] Are AGI timelines ig­nored in EA work on other cause ar­eas?

freedomandutility18 Aug 2022 12:13 UTC
20 points
6 comments1 min readEA link

Pod­cast: Mag­nus Vind­ing on re­duc­ing suffer­ing, why AI progress is likely to be grad­ual and dis­tributed and how to rea­son about poli­tics

Gus Docker21 Nov 2021 15:29 UTC
26 points
0 comments1 min readEA link
(www.utilitarianpodcast.com)

Es­ti­mat­ing the Cur­rent and Fu­ture Num­ber of AI Safety Researchers

Stephen McAleese28 Sep 2022 20:58 UTC
64 points
34 comments9 min readEA link

What if we don’t need a “Hard Left Turn” to reach AGI?

Eigengender15 Jul 2022 9:49 UTC
39 points
7 comments4 min readEA link

Open Philan­thropy’s AI grants

Vasco Grilo30 Jul 2022 17:22 UTC
21 points
2 comments2 min readEA link

GiveWell should use shorter TAI timelines

OscarD27 Oct 2022 6:59 UTC
52 points
6 comments3 min readEA link

Graph­i­cal Rep­re­sen­ta­tions of Paul Chris­ti­ano’s Doom Model

Nathan Young7 May 2023 13:03 UTC
48 points
2 comments1 min readEA link

Trans­for­ma­tive AI and Com­pute [Sum­mary]

lennart23 Sep 2021 13:53 UTC
60 points
5 comments9 min readEA link

Red-team­ing Holden Karnofsky’s AI timelines

Vasco Grilo25 Jun 2022 14:24 UTC
58 points
2 comments11 min readEA link

A con­ver­sa­tion with Ro­hin Shah

AI Impacts12 Nov 2019 1:31 UTC
27 points
8 comments33 min readEA link
(aiimpacts.org)

Rood­man’s Thoughts on Biolog­i­cal Anchors

lukeprog14 Sep 2022 12:23 UTC
72 points
8 comments1 min readEA link
(docs.google.com)

What is Com­pute? - Trans­for­ma­tive AI and Com­pute [1/​4]

lennart23 Sep 2021 13:54 UTC
48 points
6 comments18 min readEA link

[Question] How does one find out their AGI timelines?

Yadav7 Nov 2022 22:34 UTC
19 points
4 comments1 min readEA link

OpenAI’s new Pre­pared­ness team is hiring

leopold26 Oct 2023 20:41 UTC
85 points
13 comments1 min readEA link

Some Back­ground on Open Philan­thropy’s Views Re­gard­ing Ad­vanced Ar­tifi­cial Intelligence

Holden Karnofsky16 May 2016 13:08 UTC
3 points
0 comments29 min readEA link
(www.openphilanthropy.org)

New re­port on how much com­pu­ta­tional power it takes to match the hu­man brain (Open Philan­thropy)

Aaron Gertler15 Sep 2020 1:06 UTC
45 points
1 comment18 min readEA link
(www.openphilanthropy.org)

Com­pute Gover­nance and Con­clu­sions—Trans­for­ma­tive AI and Com­pute [3/​4]

lennart14 Oct 2021 7:55 UTC
20 points
3 comments5 min readEA link

Grokking “Fore­cast­ing TAI with biolog­i­cal an­chors”

anson6 Jun 2022 18:56 UTC
43 points
0 comments12 min readEA link

[Linkpost] The Prob­lem With The Cur­rent State of AGI Definitions

Yitz29 May 2022 17:01 UTC
7 points
0 comments3 min readEA link

Grokking “Semi-in­for­ma­tive pri­ors over AI timelines”

anson12 Jun 2022 22:15 UTC
60 points
1 comment11 min readEA link

Why AGI Timeline Re­search/​Dis­course Might Be Overrated

Miles_Brundage3 Jul 2022 8:04 UTC
120 points
27 comments10 min readEA link

[Question] What should I ask Ajeya Co­tra — se­nior re­searcher at Open Philan­thropy, and ex­pert on AI timelines and safety challenges?

Robert_Wiblin28 Oct 2022 15:28 UTC
23 points
10 comments1 min readEA link

An ex­am­i­na­tion of Me­tac­u­lus’ re­solved AI pre­dic­tions and their im­pli­ca­tions for AI timelines

Charles Dillon 20 Jul 2021 9:07 UTC
81 points
2 comments7 min readEA link

“In­tro to brain-like-AGI safety” se­ries—halfway point!

Steven Byrnes9 Mar 2022 15:21 UTC
8 points
0 comments2 min readEA link

[Question] Is there any re­search or fore­casts of how likely AI Align­ment is go­ing to be a hard vs. easy prob­lem rel­a­tive to ca­pa­bil­ities?

Jordan Arel14 Aug 2022 15:58 UTC
8 points
1 comment1 min readEA link

Cog­ni­tive sci­ence and failed AI fore­casts

Eleni_A18 Nov 2022 14:25 UTC
13 points
0 comments2 min readEA link

What role should evolu­tion­ary analo­gies play in un­der­stand­ing AI take­off speeds?

anson11 Dec 2021 1:16 UTC
12 points
0 comments42 min readEA link

Ar­tifi­cial In­tel­li­gence, Mo­ral­ity, and Sen­tience (AIMS) Sur­vey: 2021

Janet Pauketat1 Jul 2022 7:47 UTC
36 points
0 comments2 min readEA link
(www.sentienceinstitute.org)

Con­sid­er­a­tions on trans­for­ma­tive AI and ex­plo­sive growth from a semi­con­duc­tor-in­dus­try per­spec­tive

Muireall31 May 2023 1:11 UTC
23 points
1 comment2 min readEA link
(muireall.space)

Safety timelines: How long will it take to solve al­ign­ment?

Esben Kran19 Sep 2022 12:51 UTC
45 points
9 comments6 min readEA link

How Do AI Timelines Affect Ex­is­ten­tial Risk?

Stephen McAleese29 Aug 2022 17:10 UTC
2 points
0 comments23 min readEA link
(www.lesswrong.com)

“AGI timelines: ig­nore the so­cial fac­tor at their peril” (Fu­ture Fund AI Wor­ld­view Prize sub­mis­sion)

ketanrama5 Nov 2022 17:45 UTC
10 points
0 comments12 min readEA link
(trevorklee.substack.com)

2022 AI ex­pert sur­vey results

Zach Stein-Perlman4 Aug 2022 15:54 UTC
88 points
7 comments2 min readEA link
(aiimpacts.org)

Fun with +12 OOMs of Compute

kokotajlod1 Mar 2021 21:04 UTC
28 points
2 comments1 min readEA link
(www.lesswrong.com)

A moral back­lash against AI will prob­a­bly slow down AGI development

Geoffrey Miller31 May 2023 21:31 UTC
141 points
22 comments14 min readEA link

AI Timelines via Cu­mu­la­tive Op­ti­miza­tion Power: Less Long, More Short

Jake Cannell6 Oct 2022 7:06 UTC
27 points
0 comments1 min readEA link

[Link post] Paths To High-Level Ma­chine Intelligence

Daniel_Eth22 Sep 2021 2:43 UTC
23 points
0 comments1 min readEA link

[Question] Are AGI labs build­ing up im­por­tant in­tan­gibles?

Raven8 Apr 2022 18:43 UTC
9 points
6 comments1 min readEA link

An­nual AGI Bench­mark­ing Event

Metaculus26 Aug 2022 21:31 UTC
20 points
2 comments2 min readEA link
(www.metaculus.com)
No comments.