Do Degrees Still Matter in the Age of AI?

Cover Memo

This report draws on the AI@Work survey (CHIP50 Wave 38.6), a survey of 23,462 U.S. adults (weighted n = 23,485) fielded June 21–July 13, 2026 to boost statistical power on artificial-intelligence questions. All percentages are weighted to U.S. Census population targets and are population-representative; unweighted respondent counts are reported as a reliability guide only.

These are interim estimates. The AI@Work survey launched June 21, 2026 and data collection is ongoing; the figures in this report reflect responses received through July 13, 2026 and will be revised in a later release.

The report has two bases. Questions about the value of degrees and of specific majors (ai_deg, ai_deg_masters, ai_deg_phd, ai_degree_fields) were asked of the full adult sample (unweighted n ≈ 23,400). Questions about hiring practices (ai_deg_val_hire, ai_resume, ai_college_signal, ai_literacy_how) were asked only of respondents who report authority to hire or significantly influence hiring — an analytic base of roughly 5,100–5,250. Those subsample figures are weighted with the general Census weights and should be read as describing the subpopulation of workers with hiring authority, not a separately-calibrated employer frame. Each figure states its base.

A measurement note: the degree-value items use a "Much more valuable → Much less valuable" scale (plus "Not sure"). Coding direction (1 = much more valuable) was confirmed against the data — graduate-degree holders are the most likely to see degrees as still valuable, which is only coherent under this direction.

Key Takeaways

Introduction

Few anxieties about artificial intelligence are as widely felt as the fear that it will hollow out the value of a college education — that if a chatbot can write the essay, pass the exam, and draft the code, the credential that certified those skills is worth less. That worry is real in the public mind. But it is worth separating what people fear in the abstract from what the people who make hiring decisions actually do.

The AI@Work survey measures both. It asks the full public whether AI makes college, master's, and doctoral degrees more or less valuable, and which fields of study will gain or lose; and it asks the subset of workers with hiring authority how they now view degrees, whether they screen résumés with AI, and how they react to candidates who lean on AI. The gap between the two is the story: the public is ambivalent and mildly pessimistic about the degree, while hiring managers — the people whose opinions translate into offers — remain markedly more confident in it. AI is changing hiring, but not, so far, by making the diploma obsolete.

The public is uneasy — and the higher the degree, the safer it feels

Americans lean slightly toward thinking AI devalues a college degree — but they see advanced degrees as more protected.

Asked whether the rise of AI makes a college degree more or less valuable for the future, the public splits without a clear majority: 23% say more valuable, 31% say neither more nor less, 32% say less valuable, and 15% aren't sure. The modest tilt toward "less valuable" is the source of the popular narrative — but it is a tilt, not a verdict, and nearly half of Americans are neutral or unsure.

That unease eases as the degree gets more advanced. A bachelor's degree draws a net −8 points (23% more valuable minus 32% less); a master's is close to even (net −4); and a doctorate actually turns positive (27% more valuable vs. 23% less, net +3). Americans, in other words, see specialized, higher-level credentials as more resistant to AI than a general undergraduate degree.

Opinion also divides sharply by the respondent's own education — and in a revealing way. Graduate-degree holders are the only group that is net-positive on the college degree's AI-era value (+15 points more-vs-less), while every less-educated group is net-negative, from −7 among college graduates to −11 to −15 among those with some college or a high-school education. The least-educated are also the most uncertain: 29% of those with less than a high-school education say they are "not sure," versus 9% of graduate-degree holders (Table 1a). The pattern reads as self-interest as much as forecast — the people holding the most advanced credentials are the most convinced those credentials will hold their value, while those furthest from a degree are the most doubtful or unsure. This is a socioeconomic divide in opinion, keyed to the respondent's own place in the education system: on degree value, education sorts the attitude, not just the behavior.

Party divides it too, but not along the usual line. On the 7-point party scale, degree optimism is not a left-versus-right story: the most positive groups are the strongest partisans of both sides — Strong Republicans (net +1) and Strong Democrats (net −3) — while the most skeptical are the party leaners (Lean Republicans −18, Lean Democrats −19) and the most uncertain are pure independents (25% "not sure," the highest of any group; Table 1b). The same engagement-shaped pattern runs through the AI@Work parents-and-college report, where the most committed partisans of both parties are the ones AI has most moved toward college — suggesting political engagement, more than ideology, structures how people tie AI to the value of higher education. (These are descriptive weighted breakdowns, not modeled effects.)

But the people who hire still value the degree

Among workers with hiring authority, the "AI kills the degree" story finds little support.

Where the public leans mildly negative, hiring managers lean firmly positive. Asked how they view the value of a college degree when evaluating entry-level candidates after the rise of AI, 39% say it is more valuable, 40% say about the same, and only 15% say less (6% aren't sure). That is a net +24 points toward "more valuable," against the public's net −8 on a bachelor's — a roughly 30-point gap in outlook between the general public and the people who actually extend job offers. Far from discarding the diploma, the workers doing the hiring appear to treat it as a more meaningful signal, not a less meaningful one, in an AI-saturated applicant pool.

Comparative Context

That hiring managers hold the line on the degree while the public wavers is consistent with a labor market in which employers, facing larger and increasingly AI-assisted applicant pools, lean on verifiable signals to sort candidates — a pattern visible both in the screening behavior documented later in this report and in external hiring surveys (see How AI is already changing hiring).

Technical majors gain, the humanities slip

Asked which fields of study AI will make more or less valuable, Americans reward the technical and lab sciences and mark down the humanities and social sciences.

The pattern is consistent and intuitive. Engineering (42% "more valuable" vs. 26% "less") and computer science (44% vs. 29%) are seen as the biggest winners, with biology also net positive. Math and communication come out roughly even. The fields seen as losing ground are the humanities and social sciences: political science (22% more vs. 30% less), philosophy (23% vs. 30%), economics (26% vs. 32%), and English (27% vs. 29%). For every field, though, the most common answer is usually "no change" (40–48% for the non-technical majors), so this is a story of relative tilt rather than wholesale devaluation — Americans think AI raises the premium on building things and working with data, and modestly lowers it on writing and argument.

Inside the hiring process: AI screening and a premium on AI fluency

Hiring managers are already routing applicants through AI — and, if anything, they reward candidates who use it.

Two things are happening at once. First, AI has entered the screening pipeline: 57% of hiring managers say they use AI-based tools to screen or skim résumés at least somewhat, including 18% who do so "a lot"; the remaining 43% use it not much or not at all. Second, hiring managers lean toward rewarding an applicant's own AI use: asked how they'd react to a candidate who had used AI for all of their writing or coding, 42% say they'd be more likely to hire that person (14% "definitely" more), while just 21% would be less likely and 38% say it wouldn't matter. Far from treating heavy AI use as a red flag, a plurality of hiring managers read AI fluency as a skill, not a shortcut — a hint that the entry-level premium is shifting toward candidates who can wield the tools.

Assessment of AI skills, meanwhile, remains improvised. A third of hiring managers (34%) say they don't formally assess AI skills at all. Among those who do, the most common methods are simply asking candidates about their prior AI use (36%) and setting hands-on tasks or exercises (31%), followed by portfolios (26%), credentials or certifications (23%), and informal judgment in interviews (17%). Formal, standardized evaluation of AI ability is still the exception.

How AI is already changing hiring

Roughly half of hiring managers say AI has already changed how they make hiring decisions — and the change is concentrated among the managers who use the technology themselves.

This is not just a matter of attitude. Asked directly whether AI has changed how they make hiring decisions, 49% of hiring managers say yes — 14% "a lot" and 35% "somewhat" — while 51% say not much or not at all. And the shift maps almost entirely onto the manager's own engagement with AI: among hiring managers who never use AI in their own work, only 13% say AI has changed how they hire; among those who use it most frequently, 76% do. The people remaking hiring are the people already living inside the tools.

Where hiring has changed, the pressure lands hardest on the entry level. Among the managers who report a change (an analytic base of roughly 2,560), 27% say they are hiring fewer entry-level workers because of AI, and another 29% are keeping the same number but expecting more advanced, higher-level work from them — so a majority are either shrinking the entry rung or raising the bar to stand on it. Just 37% report no real change in their entry-level strategy, and only 7% are hiring more entry-level workers. Whether this reflects genuine substitution of AI for junior tasks or simply higher expectations, the entry-level opening is getting narrower where AI has taken hold.

The market for AI skills, meanwhile, is still loosely defined. Asked about competition for AI-capable candidates, 44% of hiring managers say they don't specifically look for AI skills at all; among those who do, 27% find such candidates hard to find and 29% find them easy. Combined with the ad-hoc assessment picture above, the impression is of a hiring system that has changed its behavior faster than it has settled on what, exactly, it is screening for.

The squeeze is visible from the workforce side too, not just from managers. Across all workers — a broader base than the hiring subsample — 37% say AI has raised the skills or experience expected at their level (12% "much higher," 24% "somewhat higher"), while 52% report no change. On job openings, workers are more circumspect: 16% say AI has reduced the number of openings at their level, 9% say it has increased them, and the plurality (41%) haven't noticed a change, with another 24% saying openings are about the same. The perception of a rising skill bar is thus considerably more widespread than the perception of outright job loss — consistent with the report's broader theme that AI is reshaping the terms of entry-level work more than it is erasing the jobs outright.

Comparative Context

External hiring surveys corroborate the screening story. A January–February 2025 Resume Genius survey of 1,000 U.S. hiring managers found 48% use AI to screen résumés before a human review — squarely in line with the 57% of surveyed hiring managers who report using AI to screen or skim résumés at least somewhat (the survey's item uses a broader "at least somewhat" threshold). A separate October 2024 Insight Global survey of 1,005 hiring managers found that 99% now use AI somewhere in the hiring process. The direction is unambiguous across independent samples: AI has moved from the margins into the everyday machinery of hiring.

Conclusion

The fear that AI is quietly cancelling the college degree is, on this evidence, more a public anxiety than a hiring reality. Americans as a whole are ambivalent and mildly pessimistic about what AI does to the value of a bachelor's, though they see advanced degrees and technical majors as more protected. The workers who actually make hiring decisions are more confident still: a plurality say a degree is more valuable in the AI era, not less, and only a small minority say less. What is changing is the mechanics around the credential — AI now helps read the résumés, candidates who use AI fluently are, if anything, rewarded for it, and about half of managers say the technology has already altered how they hire — rather than the standing of the credential itself. The sharper edge is at the entry level: where AI has taken hold, managers are more likely to shrink the junior rung or raise what they expect from it, and a third of all workers already sense a higher skills bar at their own level.

For students and workers weighing what to study, the signal is less "abandon the degree" than "the premium is shifting." Technical and quantitative fields are seen as gaining, the humanities and social sciences as modestly slipping, and — across the board — the ability to use AI well is becoming something employers look for, even if few yet know how to measure it.

Appendix A — Methods

Data source. AI@Work survey — CHIP50 Wave 38.6 (Civic Health and Institutions Project / COVID States Project). Total unweighted n = 23,462; weighted n = 23,485. Fielded June 21–July 13, 2026, with an oversample of students and adults 18–22; it is nonetheless a broad adult sample. Data collection was ongoing at the time of this extract — figures reflect responses through July 13, 2026 and are preliminary. Estimates are weighted to U.S. Census population targets.

Bases. The report uses three bases. (1) Degree-value items (ai_deg, ai_deg_masters, ai_deg_phd) and the degree-field battery (ai_degree_fields, nine majors) were asked of the full sample (unweighted n ≈ 23,400). (2) Hiring-practice items (ai_deg_val_hire, ai_resume, ai_college_signal, ai_literacy_how, hiring_decisions, hiring_compete) were asked only of respondents with authority to hire or significantly influence hiring (unweighted n ≈ 5,100–5,250); the entry-level follow-up hiring_change was asked only of that subgroup whose hiring had changed (unweighted n ≈ 2,557). (3) The two "at your level" workforce items (indr_level_quantity, indr_level_quality) were asked of all workers (unweighted n ≈ 12,400). Subsample estimates use the general Census weights and describe each subpopulation descriptively, not as a separately-calibrated frame. Every figure and table states its base.

Coding. The degree-value items are coded 1 = much more valuable … 5 = much less valuable, 6 = not sure — direction confirmed against the data (the weighted mean falls monotonically as education rises, i.e., more-educated respondents see degrees as more valuable). The degree-field battery is coded 1 = more valuable, 2 = no change, 3 = less valuable (verified against the questionnaire). ai_college_signal runs 5 = definitely encourage … 1 = definitely discourage and ai_resume 4 = a lot … 1 = not at all (0 = not sure) — both directions confirmed against the questionnaire codebook and the data (each rises monotonically with the manager's own AI-use frequency); ai_literacy_how is select-all (1 = selected). The hiring-behavior items are coded per the authoritative codebook: hiring_decisions 4 = yes a lot … 1 = not at all (direction confirmed — its weighted mean rises monotonically with the manager's own AI-use frequency, from 1.54 among non-users to 2.96 among daily users); indr_level_quality 1 = much higher AI skills expected … 3 = no change, 4 = not sure (direction confirmed — the "much higher" share rises from 3% among worker-non-users to 34% among daily users). hiring_change (1 = fewer entry-level; 2 = same #, same expectations; 3 = same #, more advanced work expected; 4 = more entry-level), hiring_compete (1 = hard to find AI-skilled; 2 = easy; 3 = don't specifically look), and indr_level_quantity (1 = fewer openings; 2 = about the same; 3 = more; 4 = not noticed; 5 = not sure) are nominal category sets, not magnitude scales, and carry no direction-reversal risk; they are reported categorically per the codebook.

Estimation. Cross-tabulations are weighted shares; no regression is reported (this is a topline descriptive report). The manager-AI-use gradients cited for hiring_decisions and indr_level_quality are descriptive weighted breakdowns by ai_work frequency, not modeled effects. this report adds descriptive breakdowns of the college-degree-value item (ai_deg) by education_cat and by the 7-point party scale (party7) — Tables 1a and 1b; party is shown on the 7-point scale because the divide is an engagement gradient (leaners and pure independents most skeptical/unsure) that the 3-point measure flattens. Select-all items are not summed. Cells with unweighted n < 10 are suppressed.

Reproducibility. Figures map to Appendix B tables; values copied verbatim from the underlying tabulations. Wording and skip logic sourced from the Qualtrics instrument via the project's authoritative AI@Work survey codebook. External comparison figures in How AI is already changing hiring were fetched and verified from the cited hiring surveys (see Appendix C).

Appendix B — Data tables

The 3-point party measure collapses this into a near-flat "everyone's slightly negative"; the 7-point scale shows it is the leaners and pure independents, not either party's base, who are most skeptical or unsure — an engagement gradient, not an ideological one.

Changed (a lot + somewhat) = 48.8%. Gradient by manager's own AI-use frequency (ai_work): % who say hiring changed rises from 12.6% among managers who never use AI at work to 76.3% among the most frequent users.

Higher skills expected (much + somewhat) = 36.5%.

Appendix C — Question wording (verbatim)

• ai_deg / ai_deg_masters / ai_deg_phd — "In your opinion, does the recent rise of AI make a college degree / master's degree / doctoral degree (Ph.D.) more or less valuable for success in the future?" (Much more valuable; Somewhat more valuable; Neither more nor less valuable; Somewhat less valuable; Much less valuable; Not sure)
• ai_degree_fields — "Now thinking about specific degree fields, do you think the following undergraduate degrees will be more or less valuable due to AI?" (More valuable; No change; Less valuable) — Economics, Math, Computer Science, Political Science, Communication, English, Philosophy, Engineering, Biology
• ai_deg_val_hire (hiring authority) — "After the recent rise of AI, how do you view the value of a college degree when evaluating entry-level candidates?" (Much more valuable → Much less valuable; Not sure)
• ai_resume (hiring authority) — "To what extent do you use AI-based tools to screen or skim resumes/CVs during the hiring process? (e.g., AI tools that summarize, rank, or extract information from resumes)" (A lot; Some; Not much; Not at all; Not sure)
• ai_college_signal (hiring authority) — "If you knew that a candidate had used AI to assist with all of their writing or coding assignments, how would this affect your decision to hire this person?" (Definitely encourage → Definitely discourage hiring)
• ai_literacy_how (hiring authority; select all) — "When hiring for entry-level roles, how do you typically assess AI skills?" (We do not assess AI skills; By asking candidates about their prior AI use; Through hands-on tasks or exercises; Portfolio or work samples; Credentials or certifications; Informal judgment during interviews; Other)
• hiring_decisions (hiring authority) — "Has AI changed how you make hiring decisions?" (Yes, a lot; Yes, somewhat; Not much; Not at all)
• hiring_change (hiring authority whose hiring changed) — entry-level hiring strategy under AI (Fewer entry-level roles because of AI; Same number, same expectations; Same number, but expect more/higher-level work with AI; More entry-level roles because of AI)
• hiring_compete (hiring authority) — market for AI-skilled candidates (Hard to find AI-skilled candidates; Easy to find; Do not specifically look for AI skills)
• indr_level_quantity (all workers) — "Has AI affected the number of job openings at your level?" (Fewer; About the same; More; Have not noticed a change; Not sure)
• indr_level_quality (all workers) — "Has AI changed the skills or experience expected at your level?" (Much higher AI skills; Somewhat higher; Have not noticed a change; Not sure)

External sources cited (Comparative Context, How AI is already changing hiring):

• Resume Genius, AI's Impact on Hiring 2025 (survey of 1,000 U.S. hiring managers, Jan 8–Feb 14 2025): 48% use AI to screen résumés before human review. https://resumegenius.com/blog/job-hunting/ai-impact-on-hiring
• Insight Global, 2025 AI in Hiring Report (survey of 1,005 hiring managers, Oct 17–22 2024): 99% use AI somewhere in the hiring process. https://insightglobal.com/2025-ai-in-hiring-report/