Pith Number

pith:DPXW2N6D

pith:2020:DPXW2N6DR3BZQPDQQKSXA3WKDN

not attested not anchored not stored refs resolved

Artificial Intelligence, Values and Alignment

Iason Gabriel

The central task for AI alignment is to identify fair principles that gain reflective endorsement from people with differing moral beliefs, rather than discovering true moral principles.

arxiv:2001.09768 v2 · 2020-01-13 · cs.CY

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{DPXW2N6DR3BZQPDQQKSXA3WKDN}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

The central challenge for theorists is not to identify 'true' moral principles for AI; rather, it is to identify fair principles for alignment, that receive reflective endorsement despite widespread variation in people's moral beliefs.

C2weakest assumption

That fair principles for AI alignment can be identified through methods like reflective endorsement or other procedures in a way that is robust to moral pluralism and sufficient to guide technical alignment work.

C3one line summary

AI alignment should target fair principles that receive reflective endorsement despite moral variation, rather than identifying true moral principles, with a principle-based approach combining different alignment elements.

References

12 extracted · 12 resolved · 2 Pith anchors

[1] Abbeel, P. & Ng, A.Y. (2004, July). Apprenticeship learning via inverse reinforcement learning. In Pro- ceedings of the twenty-first international conference on Machine learning (p. 1). ACM. Achiam, J 2004

[2] Baum, S.D. (2017). Social choice ethics in artificial intelligence. AI Soc (pp. 1–12). Beauchamp, T. L., & Childress, J. F. (2001). Principles of biomedical ethics. USA: Oxford University Press. Black 2017

[3] Cohen, G. A. (2003). Facts and principles. Philosophy & Public Affairs, 31(3), 211–245. Cohen, J. (2010). The arc of the moral universe and other essays. New York: Harvard University Press. Cohen, J., 2003

[4] Impossibility and uncertainty theorems in AI value alignment 1981 · arXiv:1901.00064

[5] Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., Dignum, V., et al. (2018). AI4People— an ethical framework for a good AI society: opportunities, risks, principles, and recommendat 2018

Formal links

2 machine-checked theorem links

Cited by

20 papers in Pith

Perception Gaps in Risk, Benefit, and Value Between Experts and Public Challenge Socially Accepted AI

AI of the People, by the People, for the People: A Social Choice Approach to Collective Control of Artificial Intelligence

ActivationReasoning: Logical Reasoning in Latent Activation Spaces

A Roadmap to Pluralistic Alignment

AI and My Values: User Perceptions of LLMs' Ability to Extract, Embody, and Explain Human Values from Casual Conversations

Receipt and verification

First computed	2026-05-17T23:38:14.336965Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

1bef6d37c38ec3983c7082a5706eca1b5159335b6557e6b8092be1f34868a4ae

Aliases

arxiv: 2001.09768 · arxiv_version: 2001.09768v2 · doi: 10.48550/arxiv.2001.09768 · pith_short_12: DPXW2N6DR3BZ · pith_short_16: DPXW2N6DR3BZQPDQ · pith_short_8: DPXW2N6D

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/DPXW2N6DR3BZQPDQQKSXA3WKDN \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 1bef6d37c38ec3983c7082a5706eca1b5159335b6557e6b8092be1f34868a4ae

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "80a96133082d42a89a0e18f4f54c8c81ee73d86ed0f3d7ec95d81d87a437a9f1",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CY",
    "submitted_at": "2020-01-13T10:32:16Z",
    "title_canon_sha256": "a0a4b9db8456af42a49e14a15b67794cd626f16fe748cc1733894fcb6f5f9166"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2001.09768",
    "kind": "arxiv",
    "version": 2
  }
}