By Lee Flanagan
✨ AI Summary:
- Unstructured, low-transparency interviewing—not video calls or AI tools—is the real source of hiring bias and candidate frustration.
- The ICO’s finding that claimed “human review” of AI recommendations often cannot be verified shows accountability gaps exist equally in face-to-face and automated processes.
- Hiring teams must document scoring criteria, use consistent questions across candidates, and retain evidence of each decision to survive regulatory scrutiny.
- Focus regulatory and process improvements on auditability and structured methodology rather than platform choice, or unaccountable processes will continue hiding behind whichever channel is hardest to inspect.
One rejected jobseeker put it starkly in the pages of the Guardian: “The hiring process has become so mechanised, both figuratively and literally, that it’s hard to believe that the people who end up getting hired aren’t merely the best at gaming the system.” Andy Burnham heard something similar on a recent employment-focused podcast, describing the jobseeking environment for young people as “harsh.” His target was the shift to Zoom and Microsoft Teams interviews, and whether applicants can properly convey their personal qualities without a face-to-face meeting.
The Guardian’s editorial backs him. I do not, not fully. The backlash against video and AI interviews aims at the wrong culprit. The actual weakness in modern hiring is unstructured, low-transparency interviewing, and it survives perfectly well whether a bot or a human asks the questions.
Burnham’s Complaint Is About the Screen, Not the Script
Burnham’s criticism targets the medium: video calls, prerecorded prompts, the absence of a shared room. Zoom’s executives answered in kind, pointing out that online interviewing cuts travel costs and gives applicants who cannot fund long journeys or overnight stays a chance to compete. Both sides are arguing about the pipe, not what runs through it.
The editorial gets closest to the real problem when it describes the “one-way” video interview: applicants given a fixed window to plan and deliver answers to prerecorded questions, unable to process feedback or read a face on the other end. That format is genuinely disorientating: speaking into a void, with no one there to respond. Hidden criteria, unexplained scoring, and no chance to clarify a bad question cause that disorientation, not the camera.
Put the identical prompts and the identical scoring gaps into a face-to-face panel interview, and the candidate leaves just as confused about why they failed.
The ICO Found the Gap That Actually Matters
Buried well below Burnham’s podcast comments is a finding that should worry hiring leaders more. The UK’s Information Commissioner’s Office is tasked by parliament with producing a code of practice on AI and automated decision-making. It found that when companies claimed AI-generated hiring recommendations were subject to human review, that claim was often only partially true, if at all.
Look at what that finding actually says. It describes accountability that could not be verified, a claimed “human in the loop” that turned out, on inspection, to be mostly loop. Fairness comes from a process that can show its evidence, whoever or whatever produced the recommendation. A human sitting in the room proves nothing by itself.
The EEOC Case Undercuts the Editorial’s Own Argument
The editorial also raises a settled US Equal Employment Opportunity Commission case: a company’s age-discrimination bias was exposed only when a rejected applicant reapplied, changing nothing but her age, and passed through to the next round. It is offered as evidence against algorithmic screening. Read closely, it is evidence against something else: opacity that only an outsider gaming the exact same system happened to catch.
The Guardian’s own editorial concedes the point one paragraph later: “It could hardly be claimed that traditional recruitment processes were bias-free.” Swap the algorithm for a hiring manager running on gut feel and undocumented criteria, and the failure mode stays the same: no scorecard, and no record a rejected applicant could ever use to prove anything. In our view, the EEOC case illustrates that an unaccountable process enables discrimination, not that algorithms are uniquely biased. Algorithms are simply the version currently making headlines.
Why the Wrong Diagnosis Will Cost TA Teams
The stakes of misdiagnosis are not abstract. Almost half of UK jobseekers have pitched to an AI bot, according to recent research cited by the Guardian, and many gave up in frustration. The Institute for Public Policy Research has found that young people in England are increasingly “losing faith in their futures.” Regulators and politicians are already looking for something to fix.
If TA leaders let this debate stay framed as AI against human, the fix that follows will target the channel: more restrictions on one-way video, more calls pushed back onto Zoom, a warm body mandated on the other end of a camera. Unstructured questions and unverifiable scoring, the actual source of unfair outcomes, will stay untouched. Ask your own process the same question the ICO asked of employers: can you prove review happened, not just that a person was technically present. Regulators should be pushed toward that standard instead of a channel mandate, because it is harder to dodge than asking whether a human asked the question.
Doing nothing costs more than a bad headline. It means the next EEOC-style case, or the next ICO audit, lands on whichever hiring process happens to be easiest to inspect, not on the one that is actually broken. A structured process, tested, scored and documented, survives that scrutiny whether it runs on Zoom, in a conference room, or through a chatbot. An unstructured one survives neither.
Original reporting: The Guardian.
Frequently asked questions
Why does the ICO’s finding matter more than the debate over Zoom interviews?
Because it exposes a gap in accountability, not a flaw in any particular channel. The ICO found that firms claiming human review of AI hiring recommendations often could not show that review had actually happened, and that same gap can hide inside a face-to-face process just as easily.
Does the EEOC age-discrimination case show that algorithms are more biased than human recruiters?
No. The Guardian’s own editorial admits that traditional recruitment was never bias-free, and the case only surfaced because a rejected applicant happened to reapply and expose it. The lesson concerns unaccountable process, not algorithms specifically.
What makes a one-way video interview different from a live one, and why does it matter?
In a one-way interview, applicants record answers to prerecorded questions with no one to respond in real time, an experience the Guardian’s editorial itself calls disorientating. That format makes an existing problem, unclear scoring and no feedback, harder to hide, though it did not create the problem.
What should hiring teams take from this instead of just changing platforms?
Focus on whether the process can produce evidence: consistent questions, documented criteria for each score, and a clear reason behind every rejection. A process that can show that evidence holds up under regulatory scrutiny, whether it runs on video, in person, or through a chatbot.