pith. sign in

arxiv: 2605.23177 · v1 · pith:6KRLNPXInew · submitted 2026-05-22 · 💻 cs.CY · cs.HC

Cognitive offloading and the speedup illusion in human-AI interaction

Sunny Yu , Myra Cheng , Ahmad Jabbar , Ilia Sucholutsky , Katherine M. Collins , Dan Jurafsky , Robert D. Hawkins This is my paper

Pith reviewed 2026-05-25 03:24 UTC · model grok-4.3

classification 💻 cs.CY cs.HC
keywords cognitive offloadingspeedup illusionhuman-AI interactiontime estimationbehavioral studyLLM productivityeffort perceptiondecision calibration
0
0 comments X

The pith

People expect AI to speed up simple cognitive tasks but actual completion times show no gain over working alone.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tests whether users can correctly forecast time savings when offloading cognitive work to AI on simple tasks. It reports that participants accurately predict their own solo completion times yet significantly underestimate times when AI is available, even though measured times turn out identical in both conditions. The same over-optimism does not appear when participants imagine receiving help from another human. Participants also rate their effort as lower with AI despite the unchanged duration. These mismatches matter because decisions about when to use AI rest on expectations of efficiency that the data show are systematically biased.

Core claim

The central claim is that a speedup illusion exists: participants hold accurate forecasts of independent completion times but systematically underestimate AI-assisted times, with no corresponding bias for imagined human assistance; actual measured completion times do not differ between conditions, while subjective effort is reported as lower with AI.

What carries the argument

The speedup illusion: the specific bias in which independent time forecasts are accurate but AI-assisted time forecasts are too low.

If this is right

  • Completion time alone does not fully characterize efficiency gains from AI, because effort perceptions dissociate from duration.
  • Calibration of time expectations is required for effective decisions about cognitive offloading to AI.
  • The bias is AI-specific rather than a general over-optimism about any external help.
  • Users may choose to offload tasks to AI based on an inflated sense of time savings that does not materialize.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the illusion persists across more complex tasks, productivity forecasts used in AI deployment planning would need adjustment for over-optimism.
  • Providing users with immediate feedback on actual versus predicted times could serve as a corrective mechanism.
  • The dissociation between time and effort suggests that subjective experience metrics may drive adoption more than objective duration.
  • The finding raises the question of whether similar illusions appear in other automation contexts such as software tools or search engines.

Load-bearing premise

The chosen tasks and AI-assistance conditions isolate the effect of AI versus no AI without confounds from differences in task difficulty or AI output quality.

What would settle it

A direct replication in which participants' predictions of AI-assisted completion times match the observed times within the same experimental setup.

Figures

Figures reproduced from arXiv: 2605.23177 by Ahmad Jabbar, Dan Jurafsky, Ilia Sucholutsky, Katherine M. Collins, Myra Cheng, Robert D. Hawkins, Sunny Yu.

Figure 1
Figure 1. Figure 1: Experiment setup: a) we include a prediction sample and a completion sample. In the prediction sample, participants predict the completion times for themselves and for using external assistance. In the completion sample, participants complete the task independently or with AI assistance. Example tasks: b), we show one example per category × difficulty level (8 of the 24 tasks). dependent completion to take… view at source ↗
Figure 2
Figure 2. Figure 2: There is no significant difference between the pre [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The difference between the two conditions per task in terms of time (left) and NASA-TLX (right), ordered by effect [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
read the original abstract

Large language models (LLMs) have the potential to boost human productivity by speeding up task completion -- provided users know when to offload cognitive work to them. But we do not know if users are well-calibrated in estimating these potential time savings. We conducted a preregistered large-scale behavioral study (N = 1237) to characterize mismatches between expectations and reality, with a focus on simple cognitive tasks. While actual completion times between independent completion and AI-assisted completion did not differ, participants predicted AI to be significantly faster. The same bias was not observed when imagining help from another human participant. We identify a speedup illusion where people have accurate forecasts of independent completion times but significantly underestimate AI-assisted times. Additionally, time and effort dissociate: participants reported lower subjective effort with AI despite equivalent completion times. This suggests that completion time itself is not sufficient to characterize efficiency gains.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript reports results from a preregistered behavioral experiment (N=1237) on simple cognitive tasks. Actual completion times were statistically equivalent between independent work and AI-assisted conditions, yet participants predicted significantly faster performance with AI assistance. No such prediction bias appeared when participants imagined assistance from another human. Subjective effort ratings were lower in the AI condition despite equivalent objective times. The authors interpret these patterns as evidence of a 'speedup illusion' in which people underestimate the time cost of AI-assisted work while accurately forecasting independent completion times.

Significance. If the core empirical pattern holds after clarification of the experimental controls, the work contributes a large-sample demonstration that human forecasts of AI productivity gains can be miscalibrated even when objective times show no net benefit. The dissociation between objective time and subjective effort is a secondary observation that may inform models of cognitive offloading. The preregistration and sample size are positive features for an empirical claim in this domain.

major comments (1)
  1. [Methods] Methods (AI-assisted condition): The manuscript states that actual completion times did not differ but provides insufficient detail on the LLM employed, prompt construction, whether participants could edit or verify outputs, and any controls for output quality or formatting overhead. Because the speedup illusion claim rests on the equivalence of actual times reflecting a true absence of net speedup rather than an artifact of the assistance implementation, this information is load-bearing for interpreting the null result.
minor comments (2)
  1. [Abstract] Abstract: The phrase 'simple cognitive tasks' is used without an example; adding one concrete task description would help readers evaluate the scope of the claimed illusion.
  2. [Results] Results: The statistical reporting for the prediction bias (e.g., exact effect sizes, confidence intervals, and correction for multiple comparisons) should be cross-checked against the preregistration to confirm alignment.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive review and for identifying areas where additional methodological detail will strengthen the manuscript. We address the major comment below and will revise accordingly.

read point-by-point responses

  1. Referee: [Methods] Methods (AI-assisted condition): The manuscript states that actual completion times did not differ but provides insufficient detail on the LLM employed, prompt construction, whether participants could edit or verify outputs, and any controls for output quality or formatting overhead. Because the speedup illusion claim rests on the equivalence of actual times reflecting a true absence of net speedup rather than an artifact of the assistance implementation, this information is load-bearing for interpreting the null result.

    Authors: We agree that the current description of the AI-assisted condition lacks sufficient detail to fully support interpretation of the null result on completion times. In the revised manuscript we will expand the Methods section to specify the exact LLM (model, version, and any sampling parameters), the prompt templates and how they were presented to participants, the interface features allowing viewing/editing/verification of outputs, and any procedures used to monitor or control for output quality and formatting time. These additions will clarify that the observed time equivalence is not an artifact of the implementation. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical behavioral study with direct measurements

full rationale

The paper reports a preregistered experiment (N=1237) comparing measured completion times and predictions for independent vs. AI-assisted tasks on simple cognitive work. Central claims rest on statistical contrasts between observed times, self-reported effort, and forecasts; no equations, fitted parameters, derivations, or self-citation chains are present that could reduce any result to its inputs by construction. The speedup illusion is an empirical pattern extracted from participant data rather than a modeled output.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim depends on the validity of self-reported predictions and effort ratings plus the assumption that the chosen tasks represent typical cognitive offloading scenarios; no free parameters or invented entities are introduced.

axioms (1)
  • domain assumption Standard assumptions of behavioral experiments including that participants follow instructions and that time measurements accurately reflect task completion.
    Implied by the preregistered study design comparing independent, AI-assisted, and human-assisted conditions.

pith-pipeline@v0.9.0 · 5699 in / 1194 out tokens · 28988 ms · 2026-05-25T03:24:33.533458+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

115 extracted references · 115 canonical work pages · 3 internal anchors

  1. [1]

    APACrefauthors \ 2025

    ahn2025preserving APACrefauthors Ahn, S. APACrefauthors \ 2025 . Preserving Critical Thinking in the Age of Large Language Models: The Paradox of Cognitive Load and Efficiency Preserving critical thinking in the age of large language models: The paradox of cognitive load and efficiency . The Korean Journal of Medicine 100 5 197--200

    work page 2025

  2. [2]

    , Massenkoff, M

    anthropic2026aeiv4 APACrefauthors Appel, R. , Massenkoff, M. , McCrory, P. , McCain, M. , Heller, R. , Neylon, T. \ Tamkin, A. APACrefauthors \ 2026 . Anthropic Economic Index report: economic primitives. Anthropic economic index report: economic primitives. APACrefURL https://www.anthropic.com/research/anthropic-economic-index-january-2026-report APACrefURL

    work page 2026

  3. [3]

    , Nushi, B

    bansal2019beyond APACrefauthors Bansal, G. , Nushi, B. , Kamar, E. , Lasecki, W S. , Weld, D S. \ Horvitz, E. APACrefauthors \ 2019 . Beyond accuracy: The role of mental models in human-AI team performance Beyond accuracy: The role of mental models in human-ai team performance . Proceedings of the AAAI conference on human computation and crowdsourcing Pro...

    work page 2019

  4. [4]

    , Rush, N

    becker2025measuring APACrefauthors Becker, J. , Rush, N. , Barnes, E. \ Rein, D. APACrefauthors \ 2025 07 . Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity. Measuring the impact of early-2025 ai on experienced open-source developer productivity. https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/

    work page 2025

  5. [5]

    \ Petty, R E

    cacioppo1982need APACrefauthors Cacioppo, J T. \ Petty, R E. APACrefauthors \ 1982 . The need for cognition. The need for cognition. Journal of personality and social psychology 42 1 116

    work page 1982

  6. [8]

    , Sucholutsky, I

    collins2024building APACrefauthors Collins, K M. , Sucholutsky, I. , Bhatt, U. , Chandra, K. , Wong, L. , Lee, M. Griffiths, T L. APACrefauthors \ 2024 . Building machines that learn and think with people Building machines that learn and think with people . Nature human behaviour 8 10 1851--1863

    work page 2024

  7. [9]

    , Molaei, S

    dhillon2024shaping APACrefauthors Dhillon, P S. , Molaei, S. , Li, J. , Golub, M. , Zheng, S. \ Robert, L P. APACrefauthors \ 2024 . Shaping human- AI collaboration: Varied scaffolding levels in co-writing with language models Shaping human- AI collaboration: Varied scaffolding levels in co-writing with language models . Proceedings of the 2024 CHI Confer...

    work page 2024

  8. [10]

    \ Risko, E F

    dunn2016toward APACrefauthors Dunn, T L. \ Risko, E F. APACrefauthors \ 2016 . Toward a metacognitive account of cognitive offloading Toward a metacognitive account of cognitive offloading . Cognitive Science 40 5 1080--1127

    work page 2016

  9. [11]

    , Bainbridge, W A

    fan2023drawing APACrefauthors Fan, J E. , Bainbridge, W A. , Chamberlain, R. \ Wammes, J D. APACrefauthors \ 2023 . Drawing as a versatile cognitive tool Drawing as a versatile cognitive tool . Nature Reviews Psychology 2 9 556--568

    work page 2023

  10. [12]

    fraisse1984perception APACrefauthors Fraisse, P. \ . APACrefauthors \ 1984 . Perception and estimation of time Perception and estimation of time . Annual review of psychology 35 1 1--37

    work page 1984

  11. [13]

    \ Kahneman, D

    fredrickson1993duration APACrefauthors Fredrickson, B L. \ Kahneman, D. APACrefauthors \ 1993 . Duration neglect in retrospective evaluations of affective episodes. Duration neglect in retrospective evaluations of affective episodes. Journal of personality and social psychology 65 1 45

    work page 1993

  12. [14]

    APACrefauthors \ 2025

    gerlich2025ai APACrefauthors Gerlich, M. APACrefauthors \ 2025 . AI tools in society: Impacts on cognitive offloading and the future of critical thinking AI tools in society: Impacts on cognitive offloading and the future of critical thinking . Societies 15 1 6

    work page 2025

  13. [15]

    APACrefauthors \ 2023

    grassini2023development APACrefauthors Grassini, S. APACrefauthors \ 2023 . Development and validation of the AI attitude scale (AIAS-4): a brief measure of general attitude toward artificial intelligence Development and validation of the AI attitude scale (aias-4): a brief measure of general attitude toward artificial intelligence . Frontiers in psycholo...

    work page 2023

  14. [16]

    APACrefauthors \ 2020

    griffiths2020understanding APACrefauthors Griffiths, T L. APACrefauthors \ 2020 . Understanding human intelligence through human limitations Understanding human intelligence through human limitations . Trends in Cognitive Sciences 24 11 873--883

    work page 2020

  15. [17]

    , Callaway, F

    griffiths2019doing APACrefauthors Griffiths, T L. , Callaway, F. , Chang, M B. , Grant, E. , Krueger, P M. \ Lieder, F. APACrefauthors \ 2019 . Doing more with less: meta-reasoning and meta-learning in humans and machines Doing more with less: meta-reasoning and meta-learning in humans and machines . Current Opinion in Behavioral Sciences 29 24--30

    work page 2019

  16. [19]

    \ Staveland, L E

    Hart_1986 APACrefauthors Hart, S G. \ Staveland, L E. APACrefauthors \ 1988 . Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research Development of nasa-tlx (task load index): Results of empirical and theoretical research . Advances in psychology Advances in psychology \ ( 52, \ 139--183). Elsevier

    work page 1988

  17. [20]

    , Goldstein, D G

    hofman2023steroids APACrefauthors Hofman, J. , Goldstein, D G. \ Rothschild, D. APACrefauthors \ 2023 December . A Sports Analogy for Understanding Different Ways to Use AI A sports analogy for understanding different ways to use ai . Harvard Business Review . APACrefURL https://www.microsoft.com/en-us/research/publication/a-sports-analogy-for-understandi...

    work page 2023

  18. [21]

    APACrefauthors \ 2025

    hooper2025cognitive APACrefauthors Hooper, V J. APACrefauthors \ 2025 . Cognitive offloading and the reshaping of human thought: The subtle influence of Artificial Intelligence Cognitive offloading and the reshaping of human thought: The subtle influence of artificial intelligence . Colloquia, Academic Journal of Culture and Thought Colloquia, academic jo...

    work page 2025

  19. [22]

    Measuring and mitigating overreliance to build human-compatible AI

    ibrahim2025measuring APACrefauthors Ibrahim, L. , Collins, K M. , Kim, S S Y. , Reuel, A. , Lamparth, M. , Feng, K. Bhatt, U. APACrefauthors \ 2025 . Measuring and mitigating overreliance is necessary for building human-compatible AI Measuring and mitigating overreliance is necessary for building human-compatible AI . arXiv preprint arXiv:2509.08010

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  20. [23]

    , Joseph, D

    jose2025outsourcing APACrefauthors Jose, B. , Joseph, D. , Mohan, V. , Alexander, E. , Varghese, S K. \ Roy, A. APACrefauthors \ 2025 . Outsourcing cognition: the psychological costs of AI-era convenience Outsourcing cognition: the psychological costs of ai-era convenience . Frontiers in Psychology 16 1645237

    work page 2025

  21. [24]

    , Kumar, A

    kelly2023capturing APACrefauthors Kelly, M. , Kumar, A. , Smyth, P. \ Steyvers, M. APACrefauthors \ 2023 . Capturing Humans’ mental models of AI : An item response theory approach Capturing humans’ mental models of AI : An item response theory approach . Proceedings of the 2023 ACM conference on fairness, accountability, and transparency Proceedings of th...

    work page 2023

  22. [25]

    APACrefauthors \ 2015

    koriat2015metacognition APACrefauthors Koriat, A. APACrefauthors \ 2015 . Metacognition: Decision making Processes in Self-monitoring and Self-regulation Metacognition: Decision making processes in self-monitoring and self-regulation . The Wiley Blackwell handbook of judgment and decision making 2 356--379

    work page 2015

  23. [26]

    , Mazurek, G

    korzynski2023artificial APACrefauthors Korzynski, P. , Mazurek, G. , Krzypkowska, P. \ Kurasinski, A. APACrefauthors \ 2023 . Artificial intelligence prompt engineering as a new digital competence: Analysis of generative AI technologies such as ChatGPT Artificial intelligence prompt engineering as a new digital competence: Analysis of generative AI techno...

    work page 2023

  24. [27]

    \ Griffiths, T L

    lieder2020resource APACrefauthors Lieder, F. \ Griffiths, T L. APACrefauthors \ 2020 . Resource-rational analysis: Understanding human cognition as the optimal use of limited computational resources Resource-rational analysis: Understanding human cognition as the optimal use of limited computational resources . Behavioral and brain sciences 43 e1

    work page 2020

  25. [28]

    \ Wickens, C D

    liu1994mental APACrefauthors Liu, Y. \ Wickens, C D. APACrefauthors \ 1994 . Mental workload and cognitive task automaticity: an evaluation of subjective and time estimation metrics Mental workload and cognitive task automaticity: an evaluation of subjective and time estimation metrics . Ergonomics 37 11 1843--1854

    work page 1994

  26. [29]

    \ Meck, W H

    matthews2016temporal APACrefauthors Matthews, W J. \ Meck, W H. APACrefauthors \ 2016 . Temporal cognition: Connecting subjective time to perception, attention, and memory. Temporal cognition: Connecting subjective time to perception, attention, and memory. Psychological bulletin 142 8 865

    work page 2016

  27. [30]

    \ Crockett, M J

    messeri2024artificial APACrefauthors Messeri, L. \ Crockett, M J. APACrefauthors \ 2024 . Artificial intelligence and illusions of understanding in scientific research Artificial intelligence and illusions of understanding in scientific research . Nature 627 8002 49--58

    work page 2024

  28. [32]

    \ Sandberg, K

    overgaard2012kinds APACrefauthors Overgaard, M. \ Sandberg, K. APACrefauthors \ 2012 . Kinds of access: different methods for report reveal different kinds of metacognitive access Kinds of access: different methods for report reveal different kinds of metacognitive access . Philosophical Transactions of the Royal Society B: Biological Sciences 367 1594 1287--1296

    work page 2012

  29. [33]

    \ Gilbert, S J

    risko2016cognitive APACrefauthors Risko, E F. \ Gilbert, S J. APACrefauthors \ 2016 . Cognitive offloading Cognitive offloading . Trends in cognitive sciences 20 9 676--688

    work page 2016

  30. [34]

    \ Nave, G

    shaw2026thinking APACrefauthors Shaw, S. \ Nave, G. APACrefauthors \ 2026 . Thinking—Fast, slow, and artificial: How AI is reshaping human reasoning and the rise of cognitive surrender. PsyArXiv Thinking—fast, slow, and artificial: How ai is reshaping human reasoning and the rise of cognitive surrender. psyarxiv . Preprint at https://osf. io/preprints/psy...

    work page 2026

  31. [36]

    APACrefauthors \ 2024

    shum2024generative APACrefauthors Shum, S B. APACrefauthors \ 2024 . Generative AI for Critical Analysis Practical Tools Cognitive Offloading and Human Agency Generative ai for critical analysis practical tools cognitive offloading and human agency . CEUR Workshop Proceedings. CEUR workshop proceedings

    work page 2024

  32. [37]

    , Bannert, M

    stadler2024cognitive APACrefauthors Stadler, M. , Bannert, M. \ Sailer, M. APACrefauthors \ 2024 . Cognitive ease at a cost: LLMs reduce mental effort but compromise depth in student scientific inquiry Cognitive ease at a cost: Llms reduce mental effort but compromise depth in student scientific inquiry . Computers in Human Behavior 160 108386

    work page 2024

  33. [40]

    , Schmitz, L

    wahn2023offloading APACrefauthors Wahn, B. , Schmitz, L. , Gerster, F N. \ Weiss, M. APACrefauthors \ 2023 . Offloading under cognitive load: Humans are willing to offload parts of an attentionally demanding task to an algorithm Offloading under cognitive load: Humans are willing to offload parts of an attentionally demanding task to an algorithm . Plos o...

    work page 2023

  34. [41]

    , Shao, Y

    wang2025ai APACrefauthors Wang, Z Z. , Shao, Y. , Shaikh, O. , Fried, D. , Neubig, G. \ Yang, D. APACrefauthors \ 2025 . How Do AI Agents Do Human Work? Comparing AI and Human Workflows Across Diverse Occupations How do AI agents do human work? comparing ai and human workflows across diverse occupations . arXiv preprint arXiv:2510.22780

    work page arXiv 2025

  35. [42]

    , Bian, J

    xiong2024search APACrefauthors Xiong, H. , Bian, J. , Li, Y. , Li, X. , Du, M. , Wang, S. Helal, S. APACrefauthors \ 2024 . When search engine services meet large language models: visions and challenges When search engine services meet large language models: visions and challenges . IEEE Transactions on Services Computing

    work page 2024

  36. [43]

    \ Neal, A

    yeo2008subjective APACrefauthors Yeo, G. \ Neal, A. APACrefauthors \ 2008 . Subjective cognitive effort: A model of states, traits, and time. Subjective cognitive effort: A model of states, traits, and time. Journal of Applied Psychology 93 3 617

    work page 2008

  37. [44]

    \ Kasarci, I

    yurt2024questionnaire APACrefauthors Yurt, E. \ Kasarci, I. APACrefauthors \ 2024 . A Questionnaire of Artificial Intelligence Use Motives: A contribution to investigating the connection between AI and motivation A questionnaire of artificial intelligence use motives: A contribution to investigating the connection between ai and motivation . International...

    work page 2024

  38. [45]

    , Kim, B K

    zauberman2009discounting APACrefauthors Zauberman, G. , Kim, B K. , Malkoc, S A. \ Bettman, J R. APACrefauthors \ 2009 . Discounting time and time discounting: Subjective time perception and intertemporal preferences Discounting time and time discounting: Subjective time perception and intertemporal preferences . Journal of Marketing Research 46 4 543--556

    work page 2009

  39. [46]

    2025 , month =

    Challapally, Aditya and Pease, Chris and Raskar, Ramesh and Chari, Pradyumna , title =. 2025 , month =

    work page 2025

  40. [47]

    Capturing Humans’ mental models of

    Kelly, Markelle and Kumar, Aakriti and Smyth, Padhraic and Steyvers, Mark , booktitle=. Capturing Humans’ mental models of

    work page

  41. [48]

    Proceedings of the

    Beyond accuracy: The role of mental models in human-AI team performance , author=. Proceedings of the

    work page

  42. [49]

    Trends in Cognitive Sciences , volume=

    Understanding human intelligence through human limitations , author=. Trends in Cognitive Sciences , volume=. 2020 , publisher=

    work page 2020

  43. [50]

    Aho and Jeffrey D

    Alfred V. Aho and Jeffrey D. Ullman , title =. 1972

    work page 1972

  44. [51]

    , author=

    Rational adaptation under task and processing constraints: implications for testing theories of cognition and action. , author=. Psychological review , volume=. 2009 , publisher=

    work page 2009

  45. [52]

    Publications Manual , year = "1983", publisher =

    work page 1983

  46. [53]

    Chandra and Dexter C

    Ashok K. Chandra and Dexter C. Kozen and Larry J. Stockmeyer , year = "1981", title =. doi:10.1145/322234.322243

    work page doi:10.1145/322234.322243 1981

  47. [54]

    Scalable training of

    Andrew, Galen and Gao, Jianfeng , booktitle=. Scalable training of

    work page

  48. [55]

    Dan Gusfield , title =. 1997

    work page 1997

  49. [56]

    Tetreault , title =

    Mohammad Sadegh Rasooli and Joel R. Tetreault , title =. Computing Research Repository , volume =. 2015 , url =

    work page 2015

  50. [57]

    A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =

    Ando, Rie Kubota and Zhang, Tong , Issn =. A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =. Journal of Machine Learning Research , Month = dec, Numpages =

    work page

  51. [58]

    Shaping human-

    Dhillon, Paramveer S and Molaei, Somayeh and Li, Jiaqi and Golub, Maximilian and Zheng, Shaochun and Robert, Lionel Peter , booktitle=. Shaping human-

    work page

  52. [59]

    Computers in Human Behavior , volume=

    Cognitive ease at a cost: LLMs reduce mental effort but compromise depth in student scientific inquiry , author=. Computers in Human Behavior , volume=. 2024 , publisher=

    work page 2024

  53. [60]

    Frontiers in psychology , volume=

    Development and validation of two instruments measuring intrinsic, extraneous, and germane cognitive load , author=. Frontiers in psychology , volume=. 2017 , publisher=

    work page 2017

  54. [61]

    arXiv preprint arXiv:2510.05969 , year=

    Probing the Difficulty Perception Mechanism of Large Language Models , author=. arXiv preprint arXiv:2510.05969 , year=

    work page arXiv

  55. [62]

    Task-Dependent Evaluation of LLM Output Homogenization: A Taxonomy-Guided Framework

    Llm output homogenization is task dependent , author=. arXiv preprint arXiv:2509.21267 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  56. [63]

    arXiv preprint arXiv:2507.09650 , year=

    Cultivating pluralism in algorithmic monoculture: The community alignment dataset , author=. arXiv preprint arXiv:2507.09650 , year=

    work page arXiv

  57. [64]

    arXiv preprint arXiv:2510.06124 , year=

    Taxonomy of User Needs and Actions , author=. arXiv preprint arXiv:2510.06124 , year=

    work page arXiv

  58. [65]

    Nature human behaviour , volume=

    Building machines that learn and think with people , author=. Nature human behaviour , volume=. 2024 , publisher=

    work page 2024

  59. [66]

    2025 , publisher=

    Gerlich, Michael , journal=. 2025 , publisher=

    work page 2025

  60. [67]

    and Rothschild, David , title =

    Hofman, Jake and Goldstein, Daniel G. and Rothschild, David , title =. 2023 , month =

    work page 2023

  61. [68]

    arXiv preprint arXiv:2509.18458 , year=

    CogniLoad: A Synthetic Natural Language Reasoning Benchmark With Tunable Length, Intrinsic Difficulty, and Distractor Density , author=. arXiv preprint arXiv:2509.18458 , year=

    work page arXiv

  62. [69]

    arXiv preprint arXiv:2505.12158 , year=

    The AI gap: How socioeconomic status affects language technology interactions , author=. arXiv preprint arXiv:2505.12158 , year=

    work page arXiv

  63. [70]

    arXiv preprint arXiv:2507.07935 , year=

    Working with AI: measuring the applicability of generative AI to occupations , author=. arXiv preprint arXiv:2507.07935 , year=

    work page arXiv

  64. [71]

    GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks

    GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks , author=. arXiv preprint arXiv:2510.04374 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  65. [72]

    2025 , month =

    Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity , author =. 2025 , month =

    work page 2025

  66. [73]

    arXiv preprint arXiv:2509.12886 , year=

    The LLM Already Knows: Estimating LLM-Perceived Question Difficulty via Hidden Representations , author=. arXiv preprint arXiv:2509.12886 , year=

    work page arXiv

  67. [74]

    Scientific reports , volume=

    Time pressure changes how people explore and respond to uncertainty , author=. Scientific reports , volume=. 2022 , publisher=

    work page 2022

  68. [75]

    Cognition , volume=

    Humans adaptively resolve the explore-exploit dilemma under cognitive constraints: Evidence from a multi-armed bandit task , author=. Cognition , volume=. 2022 , publisher=

    work page 2022

  69. [76]

    arXiv preprint arXiv:2510.25744 , year=

    Completion is Not Collaboration: Scaling Collaborative Effort with Agents , author=. arXiv preprint arXiv:2510.25744 , year=

    work page arXiv

  70. [77]

    Wang, Zora Zhiruo and Shao, Yijia and Shaikh, Omar and Fried, Daniel and Neubig, Graham and Yang, Diyi , journal=. How Do

    work page

  71. [78]

    arXiv preprint arXiv:2503.04761 , year=

    Which economic tasks are performed with ai? evidence from millions of claude conversations , author=. arXiv preprint arXiv:2503.04761 , year=

    work page arXiv

  72. [79]

    Collins and Sunnie S

    Lujain Ibrahim and Katherine M. Collins and Sunnie S. Y. Kim and Anka Reuel and Max Lamparth and Kevin Feng and Lama Ahmad and Prajna Soni and Alia El Kattan and Merlin Stein and Siddharth Swaroop and Ilia Sucholutsky and Andrew Strait and Q. Vera Liao and Umang Bhatt , journal=. Measuring and mitigating overreliance is necessary for building human-compatible

    work page

  73. [80]

    and Hern\'

    Oktar, Kerem and Collins, Katherine M. and Hern\'. Identifying, Evaluating, and Mitigating Risks of AI Thought Partnerships , year =. ACM AI Lett. , month = mar, keywords =. doi:10.1145/3803024 , abstract =

    work page doi:10.1145/3803024

  74. [81]

    , author=

    The need for cognition. , author=. Journal of personality and social psychology , volume=. 1982 , publisher=

    work page 1982

  75. [82]

    , author=

    Learned industriousness. , author=. Psychological review , volume=. 1992 , publisher=

    work page 1992

  76. [83]

    Current Opinion in Behavioral Sciences , volume=

    Doing more with less: meta-reasoning and meta-learning in humans and machines , author=. Current Opinion in Behavioral Sciences , volume=. 2019 , publisher=

    work page 2019

  77. [84]

    arXiv preprint arXiv:2410.05563 , year=

    Rational metareasoning for large language models , author=. arXiv preprint arXiv:2410.05563 , year=

    work page arXiv

  78. [85]

    Proceedings of the Human Factors and Ergonomics Society Annual Meeting , volume=

    Externalized and Extended Cognition: Cognitive Offloading for Human-Machine Teaming , author=. Proceedings of the Human Factors and Ergonomics Society Annual Meeting , volume=. 2024 , organization=

    work page 2024

  79. [86]

    Cognition distributed: How cognitive technology extends our minds , pages=

    Offloading cognition onto cognitive technology , author=. Cognition distributed: How cognitive technology extends our minds , pages=. 2008 , publisher=

    work page 2008

  80. [87]

    Plos one , volume=

    Offloading under cognitive load: Humans are willing to offload parts of an attentionally demanding task to an algorithm , author=. Plos one , volume=. 2023 , publisher=

    work page 2023

Showing first 80 references.