Curious and Afraid: How Students Feel About the Rise of AI

Cover Memo

This report draws on CHIP50 Wave 38.6 (total unweighted N = 23,462; fielded June 21–July 13, 2026), restricted to currently-enrolled students (school_recency = "currently enrolled", n ≈ 2,320 for the emotion items).

What was asked. Every respondent rated how well each of eleven emotions describes their feelings about the rise of AI, on a scale from 1 ("does not describe my feelings at all") to 10 ("describes my feelings very well"): Curious, Excited, Enthusiastic, Optimistic, Indifferent, Resigned, Anxious, Concerned, Scared, Angry, and Overwhelmed. These items are about AI in general, not schoolwork specifically — so this report is a portrait of the student mood toward AI that sits underneath the behavioral findings in the rest of this series.

Weighting. All means are survey-weighted to national demographic targets; unweighted N is a reliability cue only. The student sample is boosted and is not a probability sample of U.S. students. Partisanship uses the 7-point party7. Where a difference is described as meaningful, it is significant at p <.05 in the survey-weighted regressions reported in the Appendix; raw group gaps are otherwise described as descriptive.

Key Takeaways

Introduction

The national mood toward artificial intelligence is guarded. Pew Research Center finds that by mid-2025 roughly half of Americans across the political spectrum — 50% of Republicans and 51% of Democrats — said they were more concerned than excited about the growing use of AI in daily life.¹ That wariness is the backdrop against which this series has documented something apparently contradictory: currently-enrolled students have adopted AI faster and more completely than almost any other group, with two-thirds using it for coursework.

This report asks how those same students feel. If adoption tells us what students do, the emotion battery in Wave 38.6 tells us the sentiment underneath it — and the two do not line up the way the "digital natives embrace AI" cliché would predict. Students are not breezily optimistic about the technology they lean on daily. Their dominant feelings are concern and anxiety; their curiosity is real but coexists with unease; and their excitement, where it exists, is concentrated among particular groups. The picture is less "the generation that loves AI" than "the generation that can't look away from it" — engaged, apprehensive, and emotionally invested in a technology they expect to shape their working lives whether they welcome it or not.

Concerned, Anxious, and Curious: The Student Mood

Rank the eleven emotions students bring to AI and the apprehensive ones sit on top. Among currently-enrolled students, the highest-scoring feeling is Concerned (mean 6.4 on the 1–10 scale), followed by Anxious (5.8) and then Curious (5.6). Below those come Scared (5.3) and Overwhelmed (5.0). Only then do the positive, forward-leaning emotions appear — Excited (4.7), Resigned (4.6), Optimistic (4.5), and Enthusiastic (4.5) cluster at the bottom, essentially level with Indifferent (4.7) and just below Angry (4.8) (Table 1).

The shape of that ranking is the story. Setting aside curiosity — which is less a verdict than an orientation — every apprehensive emotion outranks every enthusiastic one. Students are more concerned than optimistic, more anxious than excited, more scared than enthusiastic. This is not indifference (Indifferent scores in the middle, and Resigned lower still), nor is it panic (no single emotion is off the charts). It is engaged unease: a generation paying close attention to AI, curious about what it can do, and net-apprehensive about what it means. The one bright thread is curiosity, and it is doing real work — it is almost certainly the emotional engine behind the high adoption documented elsewhere in this series. But curiosity pulls students toward the tool without making them feel good about its rise.

The Most Engaged Are Also the Most Activated

The students who use AI the most also feel the most about it. Compared with adults who last attended school more than a decade ago, currently-enrolled students score higher on nearly every emotion — positive and negative alike (Table 2). They are more curious (5.6 vs. 5.3) and more excited (4.7 vs. 4.3), yes, but the larger gaps are on the apprehensive side: students are markedly more Anxious (5.8 vs. 4.8, a full point higher) and more Angry (4.8 vs. 4.0), and somewhat more Overwhelmed (5.0 vs. 4.5) and Scared (5.3 vs. 4.9).

In other words, the emotional distance between students and their elders is not that students are more sanguine about AI — it is that they are more activated by it, and disproportionately in the direction of alarm. This is the affective counterpart to the behavioral paradox at the center of the series: the group most likely to use AI is not the group most at ease with it. Being surrounded by the technology, and being the cohort whose education and early careers it will most reshape, appears to raise the emotional stakes in both directions at once — with concern and anxiety rising fastest.

Comparative context. The national baseline is already wary — about half of U.S. adults across parties report being more concerned than excited about AI.¹ CHIP50 shows students running ahead of that baseline on the apprehensive emotions, not behind it.

On Campus, AI Optimism Has a Party

The single sharpest divide in how students feel about AI is partisan — and it runs exactly parallel to trust. Excitement about AI falls steadily from right to left: Strong-Republican students average 6.2 on Excited, versus 3.7 for Lean Democrats and roughly 4.5–4.7 for Democrats and Strong Democrats (Table 3). Curiosity shows the same tilt (Republicans ~6.5, independents and Lean Democrats ~5.2). The apprehensive emotions mirror it in reverse: Lean-Democrat students are the most Anxious (6.4), most Concerned (7.0), and most Scared (5.8), while Strong Republicans are the least anxious (5.4) and least concerned (5.5). Pure independents are a distinct case — low on curiosity and excitement alike (4.3 on Excited), the least engaged rather than the most fearful.

This is the affective face of a pattern this series has documented on attitudes and behavior: on campus, embrace of AI skews Republican, and it does so because trust in AI companies skews Republican. The companion "Partisan Puzzle" memo showed that trust in AI companies largely accounts for the partisan gap in whether students say AI helps them learn or beats a human teacher; the emotion data extends that mechanism to feeling itself. As the regression below confirms, once trust in AI companies is taken into account, most of the partisan gap in both excitement and anxiety fades — the parties differ in how they feel about AI in large part because they differ in how much they trust the firms building it.

Comparative context. Nationally, the old "Republicans are the AI skeptics" framing is out of date: Pew finds Republican and Democratic concern about AI had converged by 2025, with Republicans now more trusting of AI's governance.¹ The campus pattern — Republican students the most enthusiastic — is the leading edge of that realignment, not a contradiction of it.

A Gender Gap in Enthusiasm, Not in Fear

Men and women differ in how excited they are about AI — but not in how anxious. Among currently-enrolled students, men score more than a full point higher than women on Excited (5.3 vs. 4.2) and are similarly more enthusiastic and optimistic. Yet on Anxious the two are essentially identical (men 5.8, women 5.8), and women are only modestly higher on Scared (5.5 vs. 5.1) (Table 4).

This matters because it corrects an easy assumption. The well-documented gender gap in AI use — men adopt more and lean on AI more heavily, as the flagship report shows — is often read as men being more comfortable and women more wary. The emotion data says the comfort gap is real but one-sided: men are more excited, not less afraid. Women are just as anxious as men and about equally scared, but they bring less enthusiasm to balance it. The gender story in AI feeling is an enthusiasm gap, not a fear gap — and, as the regression shows, it is one of the few patterns here that does not dissolve once trust is accounted for.

What Predicts How a Student Feels

Behind the crosstabs, one variable dominates: trust in AI companies. In survey-weighted regressions of students' feelings on the standard demographics plus trust in AI companies (Table 5), trust is by far the strongest predictor. Each step up in trust is associated with a large increase in Excitement (coefficient +1.42 per point — the largest effect in any model here by a wide margin, and the term that carries the excitement model) and a meaningful decrease in Anxiety (−0.30 per point). Nothing else in the model comes close to trust's explanatory weight for how positively a student feels about AI.

Three further findings stand out. First, the partisan gap is largely a trust gap. In a model without trust, Lean-Democrat and Strong-Democrat students are significantly more anxious than Strong Republicans (coefficients +0.72 and +0.52); add trust in AI companies and both gaps shrink below the significance threshold (Lean Democrat to +0.48, p =.052; Strong Democrat to +0.32, n.s.). The parties feel differently about AI mostly because they trust its makers differently — the same mechanism the Partisan Puzzle memo found for attitudes.

Second, gender is independent of trust. Men remain significantly more excited than women (+0.55) even after trust is controlled — the enthusiasm gap is not a trust artifact. And it is specific to the positive emotions: gender does not significantly predict anxiety in either model.

Third, among current students, older is calmer — and the gradient is steep. Read on the survey's native age bands, anxiety falls almost monotonically with age: net of everything else, 21–30s are −0.29 to −0.35 less anxious than 18–20s, 31–40s −0.92 to −1.09, 41–50s −0.79 to −0.95, and students 51 and older −1.64 to −1.79 — the single largest demographic coefficient in either anxiety model. Excitement runs the other way and peaks in mid-life: +0.65 (21–30), +0.82 (31–40), +1.31 (41–50), then falling back to +0.61 among the 51+ group. The raw means tell the same story — Anxious 6.3 among 18–20s down to 4.5 among 51–60s; Excited 3.7 up to 6.1 among 41–50s. This is a genuine within-student age effect on feeling, and it is notable precisely because age does not predict whether or how much students use AI (the flagship's central correction). Older students use AI at the same rate as younger ones but feel markedly steadier about it: the youngest students carry the most anxiety and the least excitement.

Fourth, race is an axis of its own. Net of party, gender, age, and the rest, Black students are markedly more excited about AI (+0.77 on the 1–10 scale) and less anxious (−0.49), with Hispanic students showing the same pattern more mildly (Excited +0.38, Anxious −0.40); Asian students are not significantly different from White students on either emotion. This is visible without any controls at all — Black students average 5.1 vs. 4.5 for other students on Excited, and 5.4 vs. 5.9 on Anxious. And unlike the partisan gap, it does not vanish when trust is added: the excitement effect attenuates (to +0.41 for Black students and +0.32 for Hispanic students) but stays significant, so it is only partly a trust story. It is the affective face of the engagement pattern the flagship and "I Learn, They Cheat" document — the students who take up AI most readily also feel best about it, and Black students are prominent in both.

Conclusion

The students who have most thoroughly integrated AI into their education do not, on the whole, feel good about it. Their leading emotions are concern and anxiety; their curiosity is genuine but coexists with unease; and their excitement is real only for some — concentrated among the Republican and male students who also trust AI's makers most. Compared with older adults, students are not more relaxed about AI but more emotionally charged by it, and disproportionately toward alarm. This is the sentiment that sits beneath the adoption numbers, and it complicates the tidy story of a generation at ease with its tools: they use AI because they are curious and because it is there, not because they are confident about where it leads.

The one lever that consistently moves student feeling is trust in the companies building AI. It separates the excited from the anxious more powerfully than party, gender, income, or age, and it accounts for most of the partisan divide in mood. That points to something institutions and companies should sit with: for this generation, comfort with AI is not mainly a matter of exposure — students already have plenty — but of trust, which is low across most of the campus. The gender and age patterns are their own stories (men bring more enthusiasm, not less fear; anxiety falls steadily with age, from the 18–20s down to students in their fifties and beyond), but the central one is trust. A cohort this engaged and this anxious is not waiting to be sold on AI's capabilities, which it uses daily. It is waiting to be given a reason to believe the technology — and its makers — are on its side.

Appendix A — Methods

Data. CHIP50 Wave 38.6, fielded June 21–July 13, 2026; total unweighted N = 23,462. Restricted to currently-enrolled students (school_recency = "currently enrolled").

Base. Emotion items (ai_emotion_1–ai_emotion_11): n ≈ 2,315–2,327 currently-enrolled students per item (asked of all respondents; restricted here to current students). Regressions use n = 2,293–2,319 after list-wise deletion.

Weighting. Survey-weighted to national demographic targets; unweighted N is a reliability cue only. Not a probability sample of U.S. students. Feelings are self-reported.

Measures (verified coding). ai_emotion_1–ai_emotion_11 — "How well does each of the following describe your feelings about the rise of AI?" on a 1 = "does not describe my feelings at all" … 10 = "describes my feelings very well" scale. Items: 1 Curious, 2 Excited, 3 Enthusiastic, 4 Optimistic, 5 Indifferent, 6 Resigned, 7 Anxious, 8 Concerned, 9 Scared, 10 Angry, 11 Overwhelmed. pol_trust_ai — trust in AI companies, 1 = not at all … 4 = a lot. Partisanship: 7-point party7 (1 = Strong Republican … 7 = Strong Democrat). Age band: the survey's native age_cat_8 bins wherever the enrolled sample supports them — 18–20 / 21–30 / 31–40 / 41–50 — with the three oldest bins collapsed into 51+ because only 14 currently-enrolled respondents fall above age 60 (age_stu5).

Appendix B — Data tables

Table 1. How students feel about the rise of AI (currently-enrolled; weighted mean, 1–10)

Base ≈ 2,315–2,327 currently-enrolled students per item. Higher = the emotion describes the student's feelings about AI better. Weighted means.

Table 2. Students vs. adults long out of school (weighted mean, 1–10)

Bases: currently-enrolled n ≈ 2,320; out-of-school >10 yrs n ≈ 13,600. Students score higher on every emotion shown. Weighted means.

Table 3. Selected emotions by 7-point party (currently-enrolled; weighted mean, 1–10)

Base ≈ 2,315 currently-enrolled students (party-group unweighted n ≈ 94–646; treat Lean-Republican, n ≈ 95, as indicative). Weighted means.

Table 4. Selected emotions by gender (currently-enrolled; weighted mean, 1–10)

Bases: women n ≈ 1,505–1,513; men n ≈ 809–811. The gender gap is large on excitement, negligible on anxiety. Weighted means.

Table 5. What predicts students' feelings about AI (currently-enrolled; survey-weighted OLS)

Survey-weighted OLS among currently-enrolled students, n = 2,293–2,319. Full predictor set: party7 + gender + education + income + urban/rural + age band (+ trust in AI companies where noted). Reference categories: Strong Republican, Female, College degree, lowest income bracket, Rural (the default urbanicity reference), age 18–20. Age uses the survey's native age_cat_8 bins — 18–20 / 21–30 / 31–40 / 41–50 — with the three oldest bins collapsed into 51+ (age_stu5), because only 14 currently-enrolled respondents fall above age 60. Coefficients on the 1–10 emotion scale. *** p<.001, ** p<.01, * p<.05, n.s. = not significant. SEs model-based. Adding trust removes significance from both Democratic anxiety coefficients (Lean Democrat +0.72 → +0.48, p =.052; Strong Democrat +0.52 → +0.32, n.s.), while the gender and age effects persist. Race (added to the full model) is an independent axis: Black +0.77 / Hispanic +0.38 on Excited — attenuating to +0.41 / +0.32 once trust is added, but still significant — and Black −0.49 / Hispanic −0.40 on Anxious; Asian is not significantly different from White on either emotion.

Demographic robustness (SES, urbanicity, region). The race findings are integrated in the body ("What Predicts How a Student Feels" and the takeaways). SES: higher income and education predict more excitement, but not less anxiety. Urbanicity and Census region are null across the emotion items; the underlying regional spread is narrow to begin with — student AI adoption ranges only from 59.1% (Northeast) to 67.9% (West), no region significantly different from another (model F p =.054).

Appendix C — Question wording and response codes

Source: the Wave 38.6 Qualtrics instrument provided by the CHIP50 / COVID States team (the Wave 38.6 codebook, 2026-07-20), which is the authoritative record of item wording and stored numeric codes for this wave. Stems below are the codebook's condensed forms, not screen-verbatim text; response options and codes are verbatim. Where an item's stored code order differs from its on-screen display order, the stored codes are authoritative and are the ones used throughout this report.

• ai_emotion_1–11 — "How well does each of the following describe your feelings about the rise of AI?" on a 1–10 scale (1 = does not describe at all, 10 = describes very well). Emotions: 1 Curious · 2 Excited · 3 Enthusiastic · 4 Optimistic · 5 Indifferent · 6 Resigned · 7 Anxious · 8 Concerned · 9 Scared · 10 Angry · 11 Overwhelmed. Each is rated independently — they are not forced-choice and do not trade off.
• pol_trust_ai — trust that AI companies "do what is right" (4 = A lot; 3 = Some; 2 = Not too much; 1 = Not at all), from the institutional-trust battery.
• ai_course_extent — "How much of your schoolwork do you do with AI?" (5 = All; 4 = Most; 3 = Some; 2 = Not much; 1 = None).
• school_recency — "When did you last attend school or university?" (1 = Currently enrolled; 2 = Within the past 2 years; 3 = 3–5 years ago; 4 = 6–10 years ago; 5 = More than 10 years ago). Code 1 is the student base; code 5 is the out-of-school comparison group.
• party7 — derived 7-point party identification (1 = Strong Republican … 7 = Strong Democrat).