Speaking fear is the norm among students, not the exception.
It is week six, and forty students are timetabled to present a ten-minute talk to their seminar group. A few will speak with something close to ease. Most will get through it while their hearts pound and their notes shake, and a handful will email the night before asking to submit a recording instead, or simply not turn up. The lecturer marks each talk once, offers a few lines of written feedback, and moves on to the next cohort. Every student has now "done" a presentation. Almost none has actually practised one.
That gap between doing and practising sits at the centre of a problem most universities and colleges recognise but few have solved. Presentation skills are among the most requested graduate attributes by employers and among the most anxiety-inducing tasks we set students.
The traditional model asks the students who are most frightened to perform, once, in front of the people whose judgement they fear most. AI public speaking simulations offer a different route: private, repeatable practice with instant feedback, before the stakes are real.
Public speaking anxiety is not a fringe concern affecting a nervous minority. In a survey of students across two UK universities, Russell and Topham found that 80% reported oral presentations as a source of social anxiety affecting their learning and wellbeing — a higher proportion than for seminars (45–52%), group work (25–26%), or lectures (14–26%). In the United States, Dwyer and Davidson surveyed 815 college students and found 61% feared public speaking, which they selected more often than any other fear, including death.
The consequences reach beyond the moment of the talk. Grieve, Woodley, Hunt and McKay ran a qualitative survey of undergraduate and postgraduate students at the University of the West of England who feared public speaking, and their paper became the most-read article in the history of the Journal of Further and Higher Education, with well over 160,000 downloads.
They identified recurring themes: fear of being judged, uncertainty about content, and harsh self-evaluation. Students in that study described avoiding modules, staying silent in seminars, and carrying the dread for weeks. A national study of Finnish university students found that public speaking and presentation situations produced the highest anxiety of any learning situation students encountered.
This matters for educators because avoidance distorts assessment. When a capable student underperforms in a viva or a pitch because of nerves rather than knowledge, the mark measures the anxiety, not the learning.
8 in 10 UK students report oral presentations as a source of anxiety — a higher proportion than for seminars, group work or lectures.
(Russell and Topham)
The labour market has moved in the opposite direction to student comfort. In NACE's Job Outlook 2025 survey, verbal communication was sought by more than two-thirds of employers, and written communication by at least 70%, alongside problem-solving, which nearly 90% wanted evidence of on a graduate's résumé.
The World Economic Forum's Future of Jobs Report 2025, drawing on over 1,000 employers representing more than 14 million workers, placed human-centred capabilities including leadership and social influence among the core skills rising in importance to 2030, and named the skills gap as a leading barrier to business transformation.
The mismatch is widening. Cengage's 2025 Graduate Employability Report (based on 865 hiring managers, 698 instructors and 971 graduates) found that 48% of recent graduates felt unprepared even to apply for entry-level roles in their field, and described the tightest entry-level market in five years. LinkedIn's analysis has repeatedly placed communication at the top of its most in-demand skills list. Students are graduating into a market that prices the very skill their education makes them most afraid to practise.
Presentation skills are simultaneously the thing employers ask for most and the thing students avoid most. Closing that gap is core to what a degree is supposed to deliver.
The instinctive fix is more presentations. In practice, that runs into three hard limits.
The first is practice volume. Most students present once or twice per module because timetables cannot hold more. Skill improvement depends on repetition, and a single high-stakes attempt is the opposite of low-stakes repetition. The second is the feedback bottleneck. Van Ginkel and colleagues have shown that expert feedback outperforms peer feedback for developing presentation competence, but expert time is the scarcest resource in any department.
Forty individual feedback conversations per cohort, repeated often enough to matter, is not something most teaching teams can staff. The third is curriculum crowding. Cengage’s 2025 report found that half of educators dedicate 20% or less of their curriculum to workforce skills; subject content already fills the timetable.
Half of educators dedicate 20% or less of curriculum time to workforce skills. The timetable, not the will, is the constraint.
Cengage, 2025
The result is a system that certifies students have presented without giving them the reps and the coaching that would make them good at it. That is where simulation earns its place.
The research on virtual and AI-supported speaking practice has matured from novelty studies into controlled trials, and the direction of travel is consistent.
On anxiety, Reeves, Curran, Gleeson and Hanna published a meta-analysis pooling 11 studies and 508 participants. Virtual reality exposure therapy produced a large reduction in public speaking anxiety versus control, with an effect size of −1.39, statistically indistinguishable from the −1.41 they found for traditional in-person exposure. In plain terms, practising in front of a virtual audience worked about as well as practising in front of a real one for bringing fear down.
Virtual practice cut speaking anxiety about as much as the real thing: effect sizes of −1.39 versus −1.41.
Reeves et al., 2022
How much speaking anxiety dropped versus no practice (bigger bar = bigger drop).
Rodero and Larrea tested this directly with 100 university students, training the experimental group in VR with realistic distractors such as a ringing phone or an interrupting question. Anxiety measured through skin electrodermal activity fell sharply in the VR group — from a mean of 3.57 before training to 1.71 after — while the control group barely moved. Most participants said they would recommend the training to others. The mechanism is exposure: repeated, controlled contact with the feared situation teaches the nervous system that the catastrophe does not arrive.
Van Ginkel and colleagues also measured skill itself. They had first-year undergraduates practise a presentation in a virtual environment with system-generated feedback, and compared them against students who presented face-to-face with expert feedback. Both groups improved significantly across the components of presentation competence — cognition, behaviour and attitudes — and there was no significant difference between them. Technology-delivered feedback matched the expert.
The newest work extends this to AI. Karagöl and colleagues ran a ten-week study with 60 pre-service teachers using AI tools, and reported statistically significant reductions in public speaking anxiety and gains in academic speaking skill. A 2025 experimental study of 60 undergraduates in immersive VR found that AI-only feedback on presentations was as effective as feedback that also involved a teacher. That finding is what makes scale plausible, because the coaching does not have to be rationed.
Effective simulated practice needs more than a recorded talk and a grade. Drawing on this evidence, a few design features separate tools that build skill from tools that merely occupy time.
Practice has to be genuinely repeatable and private, so that anxious students can fail, reset, and try again without an audience of peers watching. Feedback has to be specific and behavioural, addressing pace, filler words, structure and eye contact rather than issuing a single score. It should let students work on both what they say and how they say it. And practice should be portable, so a student can rehearse on a laptop at 11pm the night before, not only in a booked VR lab.
This is the approach we take at Bodyswaps. Our Public Speaking AI roleplay lets learners deliver a talk in their own words to AI-powered virtual humans, then gives them personalised feedback on their delivery and their choices. What makes it work at curriculum level is customisability: educators can tailor the roleplay to their own context, so students practise the talk they actually need to give rather than a generic set piece, and the practice slots into an existing module instead of sitting beside it.
The body-swap moment, where the learner watches their own performance back from the audience’s perspective, is available on every device. Because our AI coaching uses large language models to evaluate learners against defined objectives and communication techniques, the feedback that used to require one-to-one staff time is available on demand, as often as a student wants to practise.
For anxious students, the point is that none of this happens in front of the class. The evidence on exposure suggests the private repetition is the active ingredient, rather than a watered-down substitute for realism.
Start by finding where speaking already carries assessment weight and where anxiety is quietly depressing marks. Vivas, pitches, seminar presentations and interview-style assessments are the obvious candidates. Introduce simulated practice as preparation for those existing moments rather than as a new assessment, so students arrive already rehearsed.
Position the tool as low-stakes practice, not a graded hurdle. The reason exposure works is that students can fail safely; grading the practice reintroduces the very fear you are trying to remove. Use the analytics to spot patterns. Our Bodyswaps Go dashboard and pedagogy approach let you see completion and confidence trends across a cohort, which helps target scarce staff time on the students who need a human conversation most.
Embed it where employability already lives. If your institution runs a careers studio, an employability module, or placement preparation, that is the natural home. Our work with higher education institutions and with further education colleges shows the strongest results when practice is woven into an existing pathway rather than bolted on. That is how Dr. Madeleine Pickles approached it at LJMU Business School, building AI roleplay practice into the HR curriculum to close the gap between what students study and what the job asks of them.
Set a benchmark and let it guide the next decision. A reasonable threshold is that a majority of a piloted cohort completes at least three practice attempts and reports higher confidence before their assessed talk. If you hit that, widen the rollout. If completion stalls below a third of the cohort, the barrier is usually access or timetabling rather than the tool, and the fix is making practice available on personal devices and naming it explicitly in the module. That barrier is solvable even in difficult circumstances: New Mexico State University worked out how to get headsets and practice to students learning remotely, and grew it into a university-wide soft skills programme.
Students are afraid of presenting and employers keep asking for it. The classroom on its own cannot supply enough repetitions or enough feedback to fix that. Simulated, AI-supported practice is the part of the system that scales: private enough for the anxious, and cheap enough in staff time to give every student the reps that a single graded talk never could. You can start a free trial or book a demo to try it with a cohort.
What are AI public speaking simulations?
They are practice tools where a student gives a talk to a virtual audience and gets instant feedback from AI. The student can try again as many times as they want. They can practise alone, so no real class is watching.
Do AI speaking tools really lower anxiety?
Yes. Studies show that practising in front of a virtual audience lowers fear about as much as practising in front of a real one. The more a student practises, the calmer they tend to feel. This works because the brain learns the scary thing is safe.
Can students practise without a VR headset?
Yes. Good tools work on a laptop, a browser, and a phone, as well as on VR headsets. A student can practise at home the night before a talk.
Will this replace teachers?
No. It gives students many more chances to practise than a teacher could mark by hand. Teachers can then spend their time on the students who need a real conversation most.
→ Higher education employability
→ Further education institutions
→ Case study: Human skills for Human Resources at LJMU Business School
→ Case study: Soft skills practice for students around the world at NMSU
Russell, G., & Topham, P. (2012). The impact of social anxiety on student learning and well-being in higher education. Journal of Mental Health, 21(4), 375–385. https://doi.org/10.3109/09638237.2012.694505
Dwyer, K. K., & Davidson, M. M. (2012). Is public speaking really more feared than death? Communication Research Reports, 29(2), 99–107. https://doi.org/10.1080/08824096.2012.667772
Grieve, R., Woodley, J., Hunt, S. E., & McKay, A. (2021). Student fears of oral presentations and public speaking in higher education: a qualitative survey. Journal of Further and Higher Education, 45(9), 1281–1293. https://doi.org/10.1080/0309877X.2021.1948509
Demographic predictors of public speaking anxiety among university students (2024). Current Psychology. https://link.springer.com/article/10.1007/s12144-024-06216-w
National Association of Colleges and Employers (2025). Job Outlook 2025. https://www.naceweb.org/talent-acquisition/candidate-selection/what-are-employers-looking-for-when-reviewing-college-students-resumes
World Economic Forum (2025). Future of Jobs Report 2025. https://www.weforum.org/publications/the-future-of-jobs-report-2025/
Cengage Group (2025). 2025 Graduate Employability Report. https://www.cengagegroup.com/news/press-releases/2025/cengage-group-2025-employability-report/
Reeves, R., Curran, D., Gleeson, A., & Hanna, D. (2022). A Meta-Analysis of the Efficacy of Virtual Reality and In Vivo Exposure Therapy as Psychological Interventions for Public Speaking Anxiety. Behavior Modification, 46(4), 937–965. https://doi.org/10.1177/0145445521991102
Rodero, E., & Larrea, O. (2022). Virtual reality with distractors to overcome public speaking anxiety in university students. Comunicar, 30(72), 85–96. https://doi.org/10.3916/C72-2022-07
van Ginkel, S., Gulikers, J., Biemans, H., Noroozi, O., Roozen, M., Bos, T., van Tilborg, R., van Halteren, M., & Mulder, M. (2019). Fostering oral presentation competence through a virtual reality-based task for delivering feedback. Computers & Education, 134, 78–97. https://doi.org/10.1016/j.compedu.2019.02.006
Karagöl, E., et al. (2025). The impact of AI applications on pre-service teachers’ public speaking anxiety and academic speaking skill in the context of oral presentations: A mixed-methods study. Teaching and Teacher Education. https://doi.org/10.1016/j.tate.2025.105220
Can AI feedback stand alone for fostering students’ presentation performance? A comparison of AI-only versus teacher-supported AI feedback (2025). Interactive Learning Environments. https://doi.org/10.1080/10494820.2025.2521338