Skip to content
Likelier

Action vs. inaction: figure for each choice

Letting AI write your school assignment or job application vs writing it yourself

If you act

Using AI to write the assignment

11%

If you don't

Writing it yourself

No direct measurement

At least one figure is an indirect estimate, not a measured regret rate. How each figure was built is shown below.


Career

Last reviewed 2026-05-28

Evidence quality 4.0/5

Eight-dimension review score against the quality rubric . Each dimension scored 1–5.

D1 Source verification
5/5
D2 Source authority & independence
4/5
D3 Regret-rate accuracy
2/5
D4 Source comparability
3/5
D5 Gilovich pattern
4/5
D6 Prose quality
4/5
D7 Caveat completeness
5/5
D8 Sample quality
5/5
Average 4.0/5
A laptop closed on a wooden desk beside a stack of handwritten notes and a fountain pen.
Indirect data: for this decision at least one figure comes from an indirect measure, such as outcomes, satisfaction or intent, or from an editorial estimate, rather than from a direct regret survey. The notes for each side below say how its figure was built.

If you act

Using AI to write the assignment

11%

No direct regret survey exists: no study asks students whether they regret using AI on an assignment. The 11% measures a concrete adverse outcome instead — ChatGPT-using job seekers denied a job once the interviewer discovered AI wrote the application (ResumeBuilder 2023). The often-cited 60% of student AI users agreeing AI harms critical thinking (RAND Dec 2025) is a belief about a general harm. It does not measure regret about a specific decision, so it is deliberately not used as the rate

US students ages 12–29 who use AI for schoolwork; separately, US adult job seekers who used ChatGPT for resumes/cover letters

December 2025 (RAND); February 2023 (ResumeBuilder)

If you don't

Writing it yourself

No direct measurement

No source states a figure for this side, so none is shown.

US students ages 12–29 who do not use AI for schoolwork

2024–2026

Figure for each choice (the measure is named on the page)

Related decisions

Semantically similar decisions — same territory, different trade-offs.

career

Drop out vs. finish degree

Figure for each choice (the measure is named on the page)

career

Retrain for AI disruption vs. wait

Figure for each choice (the measure is named on the page)

career

Private vs public university

Figure for each choice (the measure is named on the page)

career

College decision

Figure for each choice (the measure is named on the page)

career

China: accept gaokao vs. re-sit

Figure for each choice (the measure is named on the page)

career

Quitting a job

Figure for each choice (the measure is named on the page)

career

Asking for a raise

Figure for each choice (the measure is named on the page)

career

Remote vs office

Figure for each choice (the measure is named on the page)

There is no survey that asks students whether they regret using AI on an assignment, so this side has no measured regret rate. The number used here — 11% — is the cleanest genuine adverse outcome on the action side: a ResumeBuilder survey of 1,000 US job seekers found that 78% got an interview after submitting ChatGPT-written application materials and 59% were hired, but 11% were denied a job once the interviewer realised AI had written the application. The figure most often quoted in this debate — 60% of student AI users (67% of all students) agreeing that AI for schoolwork harms critical thinking, in the RAND American Youth Panel’s December 2025 wave (n=1,214 youth ages 12–29) — is deliberately not used as the rate, because it measures a belief about a general harm, not regret about a specific decision. The tell is that usage kept climbing anyway, from 48% to 62% of students in seven months even as that concern rose from 54% to 67%: students who believe AI is harmful in the abstract overwhelmingly keep using it, which is the opposite of what a regret signal would predict.

The evidence base for actual harm is stronger than the evidence base for actual regret. A field RCT by Hamsa Bastani and colleagues at Wharton, conducted with about 1,000 Turkish high-school math students, found that students given unrestricted ChatGPT-4 access during practice solved 48% more problems correctly during sessions but then scored 17% lower than the control group on a subsequent no-AI exam. The crucial follow-up finding: students did not recognise that they had learned less. If the harm is invisible to the person doing it, self-report regret will systematically understate true regret on a skill-development axis — students who lost skill cannot regret a loss they cannot see. Surveys of college applicants (foundry10, 2024) found about a third had used AI for admissions essays, with 6% having AI write the final draft; 72% of student ChatGPT users in the Intelligent.com survey agreed that using ChatGPT to write college essays counts as cheating, indicating moral discomfort that runs alongside continued use.

The inaction side is genuinely thin, and no published survey gives any percentage for regret among students who wrote assignments themselves. The largest available undergraduate study — the SERU Consortium survey of 95,000+ undergraduates across 20 research universities, published in Science in May 2026 — found that roughly a third of undergraduates do not use AI and that non-use skews lower-income, female, and racially underrepresented, with the researchers warning only, qualitatively and without a number, that these students “may fall behind in college and eventually the workplace.” No figure is shown for this side, because no source gives one, so no comparison of the two sides is made.

The 2023-to-2025 norm shift in this domain has been faster than any other decision in the dataset. ResumeBuilder’s hiring-manager survey in early 2023 found only 18% of hiring managers could identify a ChatGPT cover letter; by 2025, TopResume’s industry data show that roughly a third of hiring managers can spot AI-generated copy in under twenty seconds, 19.6% of recruiters would reject an AI application outright, and 52% accept proofreading or drafting support. Schools have shifted from blanket bans to acceptable-use policies. The figures cited here were taken in one short snapshot of a moving target; the magnitudes will move again, plausibly within the next twelve months.

Sources: action

Claim ledger

Every number below is what each source reported, with the verbatim quote we relied on and how we arrived at our figure. Click any link to verify directly.

2/7 sources independently verified verbatim against the cited source

  1. [1] RAND Corporation / Schwartz & Diliberti — More Students Use AI for Homework, and More Believe It Harms Critical Thinking: Selected Findings from the American Youth Panel
    More Students Use AI for Homework, and More Believe It Harms Critical Thinking: Selected Findings from the American Youth Panel
    Statistic
    67% of students agreed AI for schoolwork harms critical thinking in Dec 2025 (up from 54% earlier in 2025); among actual AI users the share was 60%, vs 78% among non-users; 62% of students used AI for homework, up from 48% in May 2025
    Excerpt
    “"Most of these students (60 percent) also expressed concern about using AI for school-related purposes. … The share of students in middle school grades and up who reported using AI for help with their homework increased from 48 percent in May 2025 to 62 percent in December 2025. … As of December 2025, 67 percent of students endorsed the statement 'The more students use AI for their schoolwork, the more it will harm their critical thinking skills' - up more than 10 percentage points from ten months earlier." ”
    Source data from
    2025-12-01
    Accessed
    2026-05-28
    Calculation
    RAND American Youth Panel survey of 1,214 youth ages 12–29, fielded December 2025, results published March 2026. This is a BELIEF measure, not a regret measure, and is NOT used as the headline rate. The 67% headline (60% among actual users, 78% among non-users) records agreement with the statement that using AI for schoolwork harms critical thinking — a general attitude about a societal harm, not "I regret having used AI on my own assignment." A student can believe AI is harmful in the abstract and feel no personal regret about their own use; the 14-percentage-point rise in usage (48→62%) even as concern climbed from 54% to 67% is direct revealed- preference evidence of exactly that gap. No survey has directly asked "do you regret having used AI for your schoolwork?", so this belief figure is retained only as context for why the action side is hard to measure, not as the regret rate. The headline rate is instead anchored on the ResumeBuilder 11% denied-a-job figure (a concrete adverse outcome) below.
  2. [2] ResumeBuilder.com (Pollfish panel) — 3 in 4 Job Seekers Who Used ChatGPT to Write Their Resume Got an Interview Verified
    3 in 4 Job Seekers Who Used ChatGPT to Write Their Resume Got an Interview
    Statistic
    46% of job seekers used ChatGPT for resumes/cover letters; 78% got an interview; 59% were hired; 11% were denied a job when the interviewer discovered ChatGPT use; 88% said they were 'somewhat' or 'highly' likely to continue using it
    Excerpt
    “"46% of job seekers are using ChatGPT to write their resumes and/or cover letters … 78% got an interview when using application materials written by ChatGPT 59% were hired after applying to a job using materials written by ChatGPT 11% were denied a job when the interviewer discovered they used ChatGPT … 88% of the total sample say they are ‘somewhat’ (41%) or ‘highly’ (47%) likely to continue using ChatGPT to write their job application materials in the future." ”
    Source data from
    2023-02-07
    Accessed
    2026-05-28
    Verification
    Excerpt independently re-fetched and confirmed word-for-word against the cited source during our grounding audit.
    Calculation
    ResumeBuilder.com online survey of 1,000 US job seekers, fielded February 2023 via Pollfish. Reputable-reference (industry panel) not peer-reviewed. The 11% denied-a-job figure is the strongest available concrete-consequence proxy on the action side for the job-application cohort: it is a hard adverse outcome rather than an attitude. The 88% continuation rate is a strong revealed- preference signal against high regret — most users would not report planning to repeat a behavior they regretted — but is consistent with regret being concentrated in the ~11% who paid a concrete cost. Not used as the headline rate because (a) sample is job seekers only, not the schoolwork cohort, and (b) the field date precedes mature employer detection norms. Used to anchor the lower bound of plausible action regret.
  3. [3] Bastani, Bastani, Sungu, Ge, Kabakcı, Mariman — Wharton / SSRN working paper — Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathematics
    Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathematics
    Statistic
    Students with unrestricted ChatGPT-4 access during practice performed 17% worse on a subsequent no-AI exam than the control group; the AI-assisted group solved 48% more practice problems correctly but failed to retain the skill
    Excerpt
    “"Consistent with prior work, our results show that access to GPT-4 significantly improves performance (48% improvement for GPT Base and 127% for GPT Tutor). However, we additionally find that when access is subsequently taken away, students actually perform worse than those who never had access (17% reduction for GPT Base)." ”
    Source data from
    2024-07-15
    Accessed
    2026-05-28
    Calculation
    Field RCT with ~1,000 Turkish high-school math students across three grade levels, four 90-minute practice sessions, posted to SSRN July 2024 by Wharton researchers. Used as the mechanism evidence underpinning the action-side regret framing: the harm the RAND respondents fear (skill erosion) is empirically real, even though students themselves did not perceive it. This creates the gap that makes "regret" particularly hard to measure — students who lost skill did not realize they had lost it, so self-report regret rates probably understate true regret. Not a US sample; treat the 17% effect size as directional. The paper is a working paper, not yet peer-reviewed at last revision.
  4. [4] Education Week (reporting on foundry10 research led by Jennifer Rubin) — 1 in 3 College Applicants Used AI for Essay Help. Did They Cheat? Verified
    1 in 3 College Applicants Used AI for Essay Help. Did They Cheat?
    Statistic
    About a third of 2023-24 college applicants used an AI tool for admissions-essay help; about 6% relied on AI to write the final draft
    Excerpt
    “"About a third of high school seniors who applied to college in the 2023-24 school year acknowledged using an AI tool for help in writing admissions essays, … about 6 percent of students overall—including some of Makena’s classmates, she said—relied on AI to write the final drafts of their essays instead of doing most of the writing themselves." ”
    Source data from
    2024-07-15
    Accessed
    2026-07-03
    Verification
    Excerpt independently re-fetched and confirmed word-for-word against the cited source during our grounding audit.
    Calculation
    foundry10 research led by Jennifer Rubin, reported by Education Week (July 2024). Added to ground the body-prose claim "Surveys of college applicants (foundry10, 2024) found about a third had used AI for admissions essays, with 6% having AI write the final draft," which cited foundry10 by name in the body text without a corresponding frontmatter source. Not a regret measure; used as contextual evidence for the scale of AI use in the action-side cohort.
  5. [5] Intelligent.com — Intelligent.com Survey Finds 1 in 10 College Applicants Are Using ChatGPT to Write Their College Application Essays
    Intelligent.com Survey Finds 1 in 10 College Applicants Are Using ChatGPT to Write Their College Application Essays
    Statistic
    72% of student ChatGPT users agree that using ChatGPT to write college application essays is cheating
    Excerpt
    “"72 percent of student ChatGPT users agree that using ChatGPT to write college application essays is cheating." ”
    Source data from
    2023-07-31
    Accessed
    2026-07-03
    Calculation
    Intelligent.com/Pollfish survey of 749 college-bound high school seniors and incoming college freshmen, fielded July 15-21, 2023; 303 respondents (41%) had used ChatGPT in the prior six months and answered the cheating-perception question. Added to ground the body-prose claim "72% of student ChatGPT users in the Intelligent.com survey agreed that using ChatGPT to write college essays counts as cheating," previously stated without a supporting frontmatter source. Not a regret measure; used as evidence of moral discomfort alongside continued use.
  6. [6] ResumeBuilder.com (Pollfish panel) — 82% of Hiring Managers Unable to Identify Cover Letters Written by ChatGPT
    82% of Hiring Managers Unable to Identify Cover Letters Written by ChatGPT
    Statistic
    Only 18% of hiring managers correctly identified all three cover letter introductions as human- or ChatGPT-written (82% could not)
    Excerpt
    “"However, when we asked hiring managers to review three cover letter introductions to see if they would be able to tell which was written by a human and which was written by ChatGPT, only 18% got all three right." ”
    Source data from
    2023-04-01
    Accessed
    2026-07-03 · archived copy
    Calculation
    ResumeBuilder.com/Pollfish survey of 1,000 US hiring managers, fielded March 22-26, 2023 (screened for management roles at companies with 2+ employees, $50k+ salary, frequent application reviewers). This is a different ResumeBuilder survey than the job-seeker survey cited above (same publisher, distinct sample of hiring managers rather than applicants). Added to ground the body-prose claim "ResumeBuilder's hiring-manager survey in early 2023 found only 18% of hiring managers could identify a ChatGPT cover letter," previously stated without a supporting frontmatter source.
  7. [7] TopResume — Survey: Where Employers Draw the Line on the Use of AI in Hiring
    Survey: Where Employers Draw the Line on the Use of AI in Hiring
    Statistic
    19.6% of recruiters would reject an AI-generated resume/cover letter outright; 33.5% of hiring managers can spot AI-generated copy in under 20 seconds; 52% accept AI for proofreading or drafting support
    Excerpt
    “"One in five (19.6%) recruiters would reject a candidate with an AI-generated resume or cover letter." "Over a third (33.5%) of hiring managers can spot an AI-generated resume in under twenty seconds." "Just over half (52%) say using AI for proofreading or drafting support is acceptable." ”
    Source data from
    2025-05-20
    Accessed
    2026-07-03
    Calculation
    TopResume/Pollfish survey of 600 US hiring managers, fielded May 15-16, 2025. Added to ground the body-prose claim "by 2025... TopResume's industry data show that roughly a third of hiring managers can spot AI-generated copy in under twenty seconds, 19.6% of recruiters would reject an AI application outright, and 52% accept proofreading or drafting support," previously stated without a supporting frontmatter source.

Sources: inaction

Claim ledger

Every number below is what each source reported, with the verbatim quote we relied on and how we arrived at our figure. Click any link to verify directly.

  1. [1] UC Berkeley News (reporting on SERU Consortium study published in Science) — The Largest Study of AI Use by Undergrads Is In, Revealing Disparities in Access — and in Cheating
    The Largest Study of AI Use by Undergrads Is In, Revealing Disparities in Access — and in Cheating
    Statistic
    Two-thirds of undergraduates used GenAI as of spring 2024; low-income, racially underrepresented, and female students used AI less, and researchers warned these students 'may fall behind in college and eventually the workplace'
    Excerpt
    “"low-income, racially underrepresented and female students using AI less. These students may fall behind in college and eventually the workplace because of unequal access to or practice using AI. … students from wealthier families can access advanced AI tools … students who don’t have resources may only … use free AI tools" ”
    Source data from
    2024-04-01
    Accessed
    2026-05-28
    Calculation
    SERU Consortium survey of 95,000+ undergraduates at 20 research universities, fielded spring 2024, published in Science May 2026. Used only as a qualitative opportunity-cost signal: the published finding frames non-use as a possible future disadvantage ("may fall behind in college and eventually the workplace"), but states NO percentage for regret or for concern about falling behind. No direct survey asks non-users "do you regret not having used AI?", so there is no measured inaction-side regret rate. No figure is shown for this side. RAND shows 78% of non-users believe AI is harmful, so most non-use is principled and unlikely to be regretted, and only an access-constrained minority, which the SERU authors flag, may regret it. This is the data-sparsest side of the entry.
  2. [2] RAND Corporation / Schwartz & Diliberti — More Students Use AI for Homework, and More Believe It Harms Critical Thinking: Selected Findings from the American Youth Panel
    More Students Use AI for Homework, and More Believe It Harms Critical Thinking: Selected Findings from the American Youth Panel
    Statistic
    78% of non-AI-using students believed AI for schoolwork harms critical thinking (Dec 2025); 38% of students were not using AI for homework as of Dec 2025
    Excerpt
    “"Non-AI users: 78 percent expressed concern vs. 60 percent among actual users. The percentage of middle school, high school, and college students using AI for homework rose from 48 percent to 62 percent." ”
    Source data from
    2025-12-01
    Accessed
    2026-05-28
    Calculation
    Same RAND panel. The 78% concern rate among non-users is direct evidence that abstention from AI is mostly principled, not regretful: the modal non-user thinks AI use is harmful and is therefore unlikely to regret their non-use. This points to low inaction regret overall, with only the residual access-constrained minority that the SERU demographic gap finding flags likely to regret it, not principled abstainers, and no figure is shown. The revealed-preference signal — total AI usage rising from 48% to 62% in seven months — is consistent with some inaction-side regret driving conversion, but cannot be cleanly decomposed from peer effects, capability improvements, or institutional normalization.

Caveats

Neither side has a direct regret survey — no published study asks "do you regret having used AI for your assignment?" or "do you regret having written it yourself?" The action-side rate is a disclosed proxy, not measured regret, and the inaction side has no figure. The action side is anchored on the cleanest genuine adverse outcome available: 11% of ChatGPT-using job seekers were denied a job once the interviewer discovered AI wrote the application (ResumeBuilder, Feb 2023). That figure applies only to the adult job-seeker cohort and predates mature employer detection norms. By 2025 the field had shifted toward accepting AI as a drafting tool while penalizing fully AI-generated submissions (TopResume and ResumeBuilder hiring-manager surveys). The action-side figure is therefore a directional floor, not a precise schoolwork-cohort rate. The often-quoted 60% (share of student AI users who agree AI harms critical thinking, RAND Dec 2025) is deliberately not used as the rate. It is a belief about a general societal harm and does not measure regret about a specific decision. The 14-percentage-point rise in usage (from 48% to 62%) even as that concern climbed (from 54% to 67%) is direct revealed-preference evidence that belief-in-harm and personal regret diverge. The Bastani et al. RCT (now published in PNAS, 2025) shows real skill erosion — students with unguarded ChatGPT scored 17% lower on a subsequent no-AI exam and did not recognize the loss — which means any self-report regret would understate true skill-development harm; it is a third-party outcome from a Turkish sample, used as mechanism evidence, not as a US regret rate. The inaction side is the data-sparsest in the pair: NO source gives a percentage for regret among students who wrote assignments themselves. The SERU study only warns qualitatively that AI non-users "may fall behind." No figure is shown for this side. Because the inaction side has no measurement, no gap between the two sides is established. Aggregating "school assignment" and "job application" as one decision also papers over real differences in incentives, detection norms, and stakes. The norm shift from 2023 to 2025 in this domain has been faster than any other entry in the dataset, so this pair has an unusually short shelf life.

Raw data: /api/decisions.json

Recently viewed on this device