Video calls have become a default mode of communication across nearly every part of modern life. Activities that once required being in the same room, like job interviews, doctor visits, and even court hearings, are now routinely conducted online. According to Columbia Business School Professor Melanie Brucks, these video interactions “run the gamut from very mundane to very consequential.”
As virtual formats continue to replace in-person meetings, glitches are an accepted and often dismissed part of the experience. A momentary freeze, a choppy sentence, a distorted image: most people treat them as minor inconveniences rather than meaningful disruptions.
In a new paper published in Nature, Brucks and her coauthors, Cornell University Professor Jacqueline Rifkin (who shares first-authorship) and University of Missouri–Kansas City Professor Jeff Johnson reveal a troubling story. Across 10 studies using both archival data and controlled experiments, they found that even tiny, fleeting audiovisual glitches meaningfully damage how we judge the people on the other side of the screen.
The culprit is a subtle perceptual reaction: glitches break the illusion of face-to-face human presence, creating a sense of “uncanniness” that undermines trust, likability, and ultimately consequential decisions.
Behind the Findings
Brucks and her co-authors used a multi-method approach spanning real-world conversations, controlled experiments, and high-stakes institutional settings. They began with an archival database of 1,645 natural video conversations, revealing that people who experienced glitches liked their partners less, felt less “heard,” and reported weaker social connection, even when the disruption was brief.
Next, the team conducted a series of tightly controlled experiments in telehealth, financial advising, and job interviews. By inserting short, under-one-second glitches during natural pauses in speech, they ensured that no information was lost and only the audiovisual smoothness was disrupted. Participants nevertheless rated communicators more negatively whenever glitches were present.
To dig deeper, the researchers developed a taxonomy of 10 common glitch subtypes and ran a study to assess how “uncanny” each one felt. These uncanniness ratings were later used to predict hiring evaluations. Finally, to explore real-world consequences, they analyzed transcripts from 472 parole hearings, using a custom “glitch dictionary” to identify hearings where glitches occurred. The team also conducted qualitative interviews with experienced sales professionals, revealing a widespread belief that glitches are benign, even as the data showed otherwise.
What’s Really Going On
Across all studies, a clear pattern emerged: glitches consistently harmed evaluations of communicators, even when they didn’t interfere with understanding. People trusted telehealth providers less, were less inclined to hire job candidates, and felt less willing to work with financial advisors after seeing glitchy videos. The mechanism was a feeling of “uncanniness,” a perceptual jolt that occurs when something about a human face or voice abruptly deviates from what feels natural. That eerie sensation broke the illusion of real-time human presence and spilled over into judgment.
Not all glitches were equal. The researchers found striking variation in how uncanny different distortions felt. Very short freezes—some lasting less than a second—ranked among the most unsettling, even more than multi-second freezes. The best predictor of negative evaluations wasn’t the type of glitch (audio vs. video) or how disruptive it appeared, but how unnatural it felt. Audio-loss glitches also carried an extra penalty: impaired comprehension.
Parole hearings in which glitches occurred were 12 percentage points less likely to result in parole than glitch-free hearings—despite all participants using identical institutional internet, making the glitches essentially random.
“We found that there was a significant relationship between whether or not a glitch was mentioned during the interaction and the likelihood at the end of the day that this person was granted parole. It is really scary, because that person had no control over whether a glitch was going to happen,” Brucks said.
The researchers also found that glitches only mattered when a call created a sense of direct social presence. When participants viewed a screen-share or slide deck, glitches didn’t feel uncanny and didn’t undermine evaluations—underscoring that the distortion of a human face or voice is what drives the effect.
The High Stakes of Small Glitches
There exists a hidden bias embedded in the virtual interactions that increasingly govern opportunities in work, health, and justice. While video calls are often framed as tools for enhancing access, unequal internet quality means that individuals with fewer resources are more likely to experience glitches and therefore face greater negative judgment. Over time, this can compound into fewer job offers, worse medical experiences, or unfair legal outcomes.
This raises critical questions for business leaders and policymakers. Organizations that rely on virtual interviews may need to provide access to reliable internet. Courts and hospitals may require clearer standards for when remote formats are appropriate. And as communication technologies move toward hyperrealistic digital presence through AI, VR, and holographic interfaces, glitches may become even more uncanny.
“As we're thinking about all the issues of AI and how to integrate AI, I don't think many people are thinking about the potential consequences of increasing realism, which is that it could just make things even more uncanny and uncomfortable when things go wrong, which they inevitably do,” Brucks said.
Glitches may never fully disappear, but awareness of their impact can help ensure they don’t quietly influence life-changing decisions.
