RSS

Tool AI

TagLast edit: Jan 9, 2025, 11:37 AM by Mateusz Bagiński

Tool AI is a type of Artificial Intelligence that is built to be used as a tool by the creators, rather than being an agent with its own action and goal-seeking behavior.

Generally meant to refer to AGI, tool AI is a proposed method for gaining some of the benefits of the intelligence while avoiding the dangers of having it act autonomously. It was coined by Holden Karnofsky, co-founder of GiveWell, in a critique of the Singularity Institute. Karnofsky proposed that, while he agreed that agent-based AGI was dangerous, it was an unnecessary path of development. His example of tool AI behavior was Google Maps, which uses complex algorithms and data to plot a route, but presents these results to the user instead of driving the user itself.

Eliezer Yudkowsky responded to this by enumerating several ways in which tool AI had similar difficulties in technical specification and safety. He also pointed out that it was not a common proposal among leading AGI thinkers.

See Also

External Links

Gw­ern’s “Why Tool AIs Want to Be Agent AIs: The Power of Agency”

habrykaMay 5, 2019, 5:11 AM
27 points
3 comments1 min readLW link
(www.gwern.net)

Thoughts on the Sin­gu­lar­ity In­sti­tute (SI)

HoldenKarnofskyMay 11, 2012, 4:31 AM
330 points
1,274 comments29 min readLW link

Tools want to be­come agents

Stuart_ArmstrongJul 4, 2014, 10:12 AM
24 points
81 comments1 min readLW link

Re­ply to Holden on ‘Tool AI’

Eliezer YudkowskyJun 12, 2012, 6:00 PM
152 points
356 comments17 min readLW link

Tools ver­sus agents

Stuart_ArmstrongMay 16, 2012, 1:00 PM
49 points
39 comments5 min readLW link

Solv­ing the whole AGI con­trol prob­lem, ver­sion 0.0001

Steven ByrnesApr 8, 2021, 3:14 PM
63 points
7 comments26 min readLW link

[In­tro to brain-like-AGI safety] 11. Safety ≠ al­ign­ment (but they’re close!)

Steven ByrnesApr 6, 2022, 1:39 PM
34 points
1 comment10 min readLW link

What is a Tool?

Jun 25, 2024, 11:40 PM
62 points
4 comments6 min readLW link

Agenty AGI – How Tempt­ing?

PeterMcCluskeyJul 1, 2022, 11:40 PM
22 points
3 comments5 min readLW link
(www.bayesianinvestor.com)

Yoshua Ben­gio ar­gues for tool-AI and to ban “ex­ec­u­tive-AI”

habrykaMay 9, 2023, 12:13 AM
53 points
15 comments7 min readLW link
(yoshuabengio.org)

Su­per­in­tel­li­gence 16: Tool AIs

KatjaGraceDec 30, 2014, 2:00 AM
12 points
38 comments7 min readLW link

Think­ing of tool AIs

Michele CampoloNov 20, 2019, 9:47 PM
6 points
2 comments4 min readLW link

In defense of Or­a­cle (“Tool”) AI research

Steven ByrnesAug 7, 2019, 7:14 PM
22 points
11 comments4 min readLW link

The Fu­sion Power Gen­er­a­tor Scenario

johnswentworthAug 8, 2020, 6:31 PM
155 points
30 comments3 min readLW link

Re­ply to Holden on The Sin­gu­lar­ity Institute

lukeprogJul 10, 2012, 11:20 PM
69 points
214 comments26 min readLW link

AI: re­quire­ments for per­ni­cious policies

Stuart_ArmstrongJul 17, 2015, 2:18 PM
11 points
3 comments3 min readLW link

Su­per­in­tel­li­gence 15: Or­a­cles, ge­nies and sovereigns

KatjaGraceDec 23, 2014, 2:01 AM
11 points
30 comments7 min readLW link

The Self-Unaware AI Oracle

Steven ByrnesJul 22, 2019, 7:04 PM
21 points
38 comments8 min readLW link

Paper: Iden­ti­fy­ing the Risks of LM Agents with an LM-Emu­lated Sand­box—Univer­sity of Toronto 2023 - Bench­mark con­sist­ing of 36 high-stakes tools and 144 test cases!

Singularian2501Oct 9, 2023, 12:00 AM
6 points
0 comments1 min readLW link

Some rea­sons why a pre­dic­tor wants to be a consequentialist

Lauro LangoscoApr 15, 2022, 3:02 PM
23 points
16 comments5 min readLW link

[Question] Favourite new AI pro­duc­tivity tools?

Gabe MJun 15, 2022, 1:08 AM
14 points
5 comments1 min readLW link

Deon­tol­ogy and Tool AI

Nathan1123Aug 5, 2022, 5:20 AM
4 points
5 comments6 min readLW link

In­ter­pretabil­ity/​Tool-ness/​Align­ment/​Cor­rigi­bil­ity are not Composable

johnswentworthAug 8, 2022, 6:05 PM
143 points
13 comments3 min readLW link

[Question] What is the prob­a­bil­ity that a su­per­in­tel­li­gent, sen­tient AGI is ac­tu­ally in­fea­si­ble?

Nathan1123Aug 14, 2022, 10:41 PM
−3 points
6 comments1 min readLW link

Simulators

janusSep 2, 2022, 12:45 PM
631 points
168 comments41 min readLW link8 reviews
(generative.ink)

Gen­er­a­tive, Epi­sodic Ob­jec­tives for Safe AI

Michael GlassOct 5, 2022, 11:18 PM
11 points
3 comments8 min readLW link

Ap­ply­ing su­per­in­tel­li­gence with­out col­lu­sion

Eric DrexlerNov 8, 2022, 6:08 PM
109 points
63 comments4 min readLW link

A multi-dis­ci­plinary view on AI safety research

Roman LeventovFeb 8, 2023, 4:50 PM
46 points
4 comments26 min readLW link

[Question] Re­quest for com­ments/​opinions/​ideas on safety/​ethics for use of tool AI in a large health­care sys­tem.

bokovMay 24, 2024, 8:53 PM
5 points
2 comments1 min readLW link

Cyborgism

Feb 10, 2023, 2:47 PM
339 points
46 comments35 min readLW link2 reviews

Pro­tect­ing agent boundaries

ChipmonkJan 25, 2024, 4:13 AM
11 points
6 comments2 min readLW link

[Question] Plau­si­bil­ity of cy­bor­gism for pro­tect­ing bound­aries?

ChipmonkMar 27, 2024, 6:53 PM
10 points
6 comments1 min readLW link

[Question] How to de­vour 5000 pages within a day if Chat­gpt crashes upon the +50mb file con­tain­ing the con­tent? Need some recom­men­da­tions.

GameSep 27, 2024, 7:30 AM
1 point
0 comments1 min readLW link

The Hopium Wars: the AGI En­tente Delusion

Max TegmarkOct 13, 2024, 5:00 PM
221 points
60 comments9 min readLW link

[Question] Share AI Safety Ideas: Both Crazy and Not

ankMar 1, 2025, 7:08 PM
16 points
28 comments1 min readLW link

Give Neo a Chance

ankMar 6, 2025, 1:48 AM
3 points
7 comments7 min readLW link

What’s the fu­ture of AI hard­ware?

Itay DreyfusJun 17, 2024, 1:05 PM
2 points
0 comments8 min readLW link
(productidentity.co)

An­no­tated re­ply to Ben­gio’s “AI Scien­tists: Safe and Use­ful AI?”

Roman LeventovMay 8, 2023, 9:26 PM
18 points
2 comments7 min readLW link
(yoshuabengio.org)

GPT as an “In­tel­li­gence Fork­lift.”

boazbarakMay 19, 2023, 9:15 PM
49 points
27 comments3 min readLW link

[Question] Why not tool AI?

smitheeJan 19, 2019, 10:18 PM
19 points
11 comments1 min readLW link