Personality tests aren’t corporate astrology. Weak interviews are what make them look like it.

A software developer in Gurugram says he was fired on his second day of work after a personality assessment scored him “neutral” instead of the extrovert profile the company was recruiting for, according to India Today. The employer has not confirmed the account, so treat it as an allegation, not a settled fact. The allegation has still done its job: it has reopened a fight about whether AI personality tests are useful science or corporate astrology dressed up as data. That framing asks the wrong question. The test is not the problem. The absence of anything else in the room to argue with it is.

The Verdict a Score Was Never Meant to Deliver

If the Gurugram allegation is accurate, the failure sits well upstream of the algorithm. A single psychometric label, “neutral” versus “extrovert,” apparently carried enough weight to end a two-day-old hire. This was not a rogue AI overruling human judgment. Human judgment was barely in the room to begin with. Somewhere in that hiring process, a structured interview, a work sample, a reference check should have generated a competing body of evidence heavy enough to make one behavioural label irrelevant. If none did, the score carried the entire decision by default, with nothing else built to challenge it.

Preferences, Not Performance, and That Cuts Both Ways

Ashish Dhawan, Managing Partner at NGS Global India, draws the line experts keep landing on: “The real danger is when AI and psychometric tools create a false sense of certainty. Behavioural assessments can predict preferences, not performance, resilience, integrity or the ability to learn.” Research published in Frontiers in Psychology backs the caution on the AI side specifically: AI-generated personality scores may appear statistically reliable but often fail to predict meaningful workplace outcomes.

That same limitation, predicting preference rather than performance, describes most unstructured interviews too. A hiring manager who decides someone is not a culture fit after twenty minutes of unscripted conversation is converting a surface impression into a hard verdict, the same operation as the algorithm everyone worries about. The algorithm at least shows its scoring; the gut-feel interview usually does not show its work at all. If a personality score deserves scrutiny for masquerading as certainty, so does an unstructured interview doing the same thing without anyone noticing.

When the Bias Sits Inside the Tool

Here is the complication a purely process-focused argument cannot wave away. Studies cited alongside that Frontiers in Psychology research raise concerns that AI-driven personality and behavioural assessments can disadvantage neurodivergent candidates and, in some cases, produce discriminatory outcomes, reinforcing calls for greater transparency and human oversight in hiring. That is a different failure than a hiring manager over-trusting a score: it is the score itself carrying a systematic distortion before anyone gets to weigh it against other evidence.

Our read is that this sharpens the argument rather than breaking it. An unstructured interview can be biased in ways nobody can audit, since there is no scorecard to inspect. A flawed algorithm is at least auditable: test it for adverse impact by group, demand the vendor show its validation data. Transparency and human oversight, the two remedies the research points to, are structural fixes, the same category a properly audited interview process needs. A tool capable of biased outcomes is reason to build a stronger counterweight around it, not reason to decide the counterweight does not matter.

One Rejected Report, One Strong Hire

Dhawan’s own example makes the fix explicit rather than implied. On an executive search assignment, a candidate received a negative recommendation from a personality assessment. The leadership team did not discard the report and hire on instinct. They ran multiple rounds of interviews, extensive reference checks and repeated interactions, then reached a different conclusion. “They hired him despite the report, and he went on to become one of their strongest performers. Had they relied solely on the assessment, they would have missed an outstanding hire,” Dhawan says.

Notice what did not happen. Nobody argued the assessment was junk science and should be scrapped. The organisation kept the input and added enough additional, structured evidence to outweigh it once the evidence pointed the other way. The score is one vote among several, not the only vote that counts. Dhawan’s wider point follows the same logic: “The best organisations don’t hire a personality type; they hire people who can deliver results. If every sales role demanded extroverts or every leadership role required dominant personalities, we’d lose an enormous amount of talent,” he explains.

Audit the Interview, Not the Vendor

The instinct after a story like the Gurugram allegation is to go looking for a better assessment vendor, or to drop personality testing altogether. That is the wrong remedy. The right response is to audit your own interview process. If a single psychometric label can end a candidacy in your organisation, your interviews are not generating enough independent, evidence-based signal to counterbalance it. That gap exists whether or not you use AI scoring at all. The assessment just makes it visible.

Trust your interviews, but make them earn that trust. Structured interviews with defined criteria, work samples tied to the actual job, and reference checks that ask specific questions produce evidence a single score cannot override on its own. Without that counterweight, it does not matter whether the deciding input came from an algorithm or a manager’s hunch. Either way, one data point became destiny. Ask yourself which one your last rejection actually was, and whether your process would survive the answer.

Original reporting: India Today.