A new systematic review, posted to arXiv on October 29 by Santiago Berrezueta-Guzman, Andrei Koshelev, Habiba-Loai Khalil, and Stefan Wagner, does something unusual for Apple Vision Pro coverage: it steps back from any single use case and reads the academic literature about the headset as a body of evidence, then asks what that body actually says — and doesn’t say.
The method: the authors pulled 218 records from Scopus mentioning AVP, then narrowed to 104 papers (spanning 2024–2026) with AVP as a primary focus, using explicit inclusion and exclusion criteria reported in full alongside a PRISMA-style screening flow diagram. They organize the resulting corpus using the tripartite structure of Sergio Leone’s 1966 film — the good, the bad, and the ugly — as a framing device for three categories: validated, working applications; persistent technical and ergonomic limitations of a still-early-generation headset; and ethical concerns and structural harms that follow from deliberate design, marketing, and deployment choices.
The headline finding is about the shape of the literature itself, not about the headset. Medicine and surgery, education, and software tooling together account for over half of the included papers — the “good” category dominates by sheer paper count. Documented risks or misuse are comparatively thin across the corpus. The authors are explicit that they treat this skew “as a finding in its own right rather than a limitation to be explained away” — meaning the imbalance itself is evidence about how AVP research gets funded, pitched, and published, not just a gap future researchers should fill.
Several recurring issues surface across multiple papers independently: price, simulated social presence (how AVP’s passthrough video and digital “Persona” avatars affect how real a remote conversation partner feels), and gaze-tracking privacy. The review treats these as structural rather than incidental — the same concerns keep reappearing across unrelated studies and use cases, which is itself a signal that they’re inherent to the device’s design rather than specific to any one deployment.
Why this is worth reading even if you’ve followed AVP coverage closely: individual reviews and demos tend to answer “does this work for my use case,” while a corpus-level review like this one answers a different question — what has the research community collectively chosen to study, and what has it collectively avoided? For a device three years into its public life, a literature review that foregrounds its own selection bias is more useful than another hands-on impression piece.
An independent look at where Apple Vision Pro stands today, for context alongside the review’s findings.
Related Reading
- Apple Vision Pro: The good, the bad, the ugly — arXiv:2610.40108
- Which VR Headset’s Eye Tracking Can You Actually Trust? — our related coverage on XR measurement rigor
- GraspPressure Lets You Click Through a UI While Both Hands Are Full — our coverage of another HCI-research piece
- Getting Started: Text-to-3D Generation — relevant for anyone building AVP-targeted spatial content