Pith Number

pith:J23ECVWP

pith:2026:J23ECVWPWVGWE5EOASITDNMQEA

not attested not anchored not stored refs resolved

Jobs' AI Exposure Should Be Measured from Evidence, Not Model Priors

Luca Mouchel, Pierre Bouquet, Yossi Sheffi

AI job exposure should be measured with retrieved evidence of real capabilities rather than zero-shot LLM assertions.

arxiv:2605.15474 v1 · 2026-05-14 · cs.IR

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{J23ECVWPWVGWE5EOASITDNMQEA}

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

Evidence-grounded measurement using retrieved documents better captures what current AI systems can plausibly do than zero-shot model assertions alone, as shown by higher human and automatic preference rates and closer alignment with observed real-world AI usage.

C2weakest assumption

The assumption that the retrieved news articles and academic paper abstracts constitute sufficient, representative, and unbiased evidence of current AI capabilities across all tasks, without major gaps in coverage or retrieval-induced selection effects (invoked in the description of the retrieval-augmented framework).

C3one line summary

The authors propose a retrieval-augmented framework that grounds AI exposure labels for 18,796 O*NET occupation-task pairs in retrieved news and academic abstracts, outperforming zero-shot prompting in 72% of disagreements and aligning better with observed real-world usage.

References

51 extracted · 51 resolved · 5 Pith anchors

[1] Automation and new tasks: How technology displaces and reinstates labor.Journal of Economic Perspectives, 33(2):3–30, May 2019 2019 · doi:10.1257/jep.33.2.3

[2] Autor, Frank Levy, and Richard J 2003 · doi:10.1162/003355303322552801

[3] L., Lopez, G., Olteanu, A., Sim, R., and Wallach, H 2021 · doi:10.18653/v1/2021.acl-long.81

[4] Rishi Bommasani, Scott Singer, et al 2025

[5] Measuring the intensive margin of work: Task shares and concentration 2026

Receipt and verification

First computed	2026-05-20T00:01:00.450030Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

4eb64156cfb54d62748e049131b5902027bc889ef7f341f6f771b1cba03a5695

Aliases

arxiv: 2605.15474 · arxiv_version: 2605.15474v1 · doi: 10.48550/arxiv.2605.15474 · pith_short_12: J23ECVWPWVGW · pith_short_16: J23ECVWPWVGWE5EO · pith_short_8: J23ECVWP

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/J23ECVWPWVGWE5EOASITDNMQEA \
  | 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: 4eb64156cfb54d62748e049131b5902027bc889ef7f341f6f771b1cba03a5695

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "8c622ce17b31ca673b12be8c9761f89f70de1461724b65aebc076676f5aa73fa",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.IR",
    "submitted_at": "2026-05-14T23:29:42Z",
    "title_canon_sha256": "5822a580761f7df45ea0ffd91f3cd8acf5e26ee27c37d225f15020e92f96cbba"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.15474",
    "kind": "arxiv",
    "version": 1
  }
}