pith. sign in
Pith Number

pith:GNR3B7VD

pith:2026:GNR3B7VDZSLRZHI6XVOUCXM5F3
not attested not anchored not stored refs resolved

Access Timing as Scaffolding: A Reinforcement Learning Approach to GenAI in Education

Davinia Hern\'andez-Leo, Janne Rotter, Pau Benazet i Montobbio

Strategically timed GenAI access decided by a reinforcement learning agent improves post-test performance and metacognitive accuracy over unrestricted or withheld use.

arxiv:2605.15850 v1 · 2026-05-15 · cs.CY · cs.AI · cs.HC

Add to your LaTeX paper
\usepackage{pith}
\pithnumber{GNR3B7VDZSLRZHI6XVOUCXM5F3}

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

Strategically timed GenAI access under the reinforcement learning condition improved objective post-test performance and metacognitive accuracy compared with unrestricted access, while reducing task errors and time on task relative to complete withholding, all without the need for explicit metacognitive prompts or structured scaffolding.

C2weakest assumption

The reward function derived from metacognitive theory, cognitive load theory, and productive failure produces access decisions that causally improve learning and metacognition in the absence of any explicit scaffolding.

C3one line summary

A reinforcement learning agent for timing GenAI access in education yields better post-test performance and metacognitive accuracy than unrestricted use and fewer errors than full restriction in a 105-participant lab study.

References

96 extracted · 96 resolved · 2 Pith anchors

[1] Collaborating with generative AI for learning? 2025 · doi:10.22318/cscl2025.882858
[2] Experimental evidence on the productivity effects of generative artificial intelligence 2023
[3] Promises and challenges of generative artificial intelligence for human learn- ing 2024
[4] Digital Education Council.Global AI Student Sur- vey 2024. Tech. rep. Accessed: 2025-10-27. Digi- tal Education Council, 2024.url:https://www. digitaleducationcouncil.com/post/digital- education - cou 2024
[5] AI tools in society: Impacts on cognitive offloading and the future of critical think- ing 2025

Formal links

1 machine-checked theorem link

Receipt and verification
First computed 2026-05-20T00:01:21.823950Z
Builder pith-number-builder-2026-05-17-v1
Signature Pith Ed25519 (pith-v1-2026-05) · public key
Schema pith-number/v1.0

Canonical hash

3363b0fea3cc971c9d1ebd5d415d9d2efe64eee7d8f0e6378895032cd0bee172

Aliases

arxiv: 2605.15850 · arxiv_version: 2605.15850v1 · doi: 10.48550/arxiv.2605.15850 · pith_short_12: GNR3B7VDZSLR · pith_short_16: GNR3B7VDZSLRZHI6 · pith_short_8: GNR3B7VD
Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/GNR3B7VDZSLRZHI6XVOUCXM5F3 \
  | 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: 3363b0fea3cc971c9d1ebd5d415d9d2efe64eee7d8f0e6378895032cd0bee172
Canonical record JSON
{
  "metadata": {
    "abstract_canon_sha256": "ec7f52ac70e63712355a444a1ef890421a1422929369f0d8d0e4b546fd21f587",
    "cross_cats_sorted": [
      "cs.AI",
      "cs.HC"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CY",
    "submitted_at": "2026-05-15T11:02:16Z",
    "title_canon_sha256": "7fc80e0442cfe34fee46d59064c88d9b52ca5f1ea7fa84fc98550b7ce7417996"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.15850",
    "kind": "arxiv",
    "version": 1
  }
}