Hiring Signals in the AI Era

About This Conversation

Every hiring team runs on a small handful of signals: the resume, the pedigree, the referral, and the interview. Most of us have quietly suspected for years that those signals were getting weaker. AI took a slow erosion and turned it into a collapse, and it happened fast enough that the tools never changed even though what they were measuring did. In this conversation, Taylor Sullivan and Falen Milton examine the evidence sitting underneath your most confident hiring decisions and ask whether it is anywhere near as strong as it looks. The discussion covers where each of the four signals breaks down, why adding one more verification step usually costs you the candidate you wanted, how to name a suspicion without coding it as bias, the real dollar cost of a mis-hire, and what a portable skills passport would change about the next five years of hiring.

Speakers

Taylor Sullivan is VP of Product and Assessment at Workera, where she has been pioneering AI-native skill assessments that deliver verified skills intelligence. With over fifteen years of experience as an industrial-organizational psychologist rooted in psychometrics, she has spent the bulk of her career in pre-hire assessment and has shaped modern approaches to ethical, scalable, and adaptive measurement.

Falen Milton is VP of People at Workera, where she leads people and talent. She built her career in talent acquisition before joining Workera four and a half years ago, and now works at the intersection of recruiting practice and AI adoption, navigating the same shifts her peers across the industry are facing in real time.

Speakers

Taylor Sullivan is VP of Product and Assessment at Workera, where she has been pioneering AI-native skill assessments that deliver verified skills intelligence. With over fifteen years of experience as an industrial-organizational psychologist rooted in psychometrics, she has spent the bulk of her career in pre-hire assessment and has shaped modern approaches to ethical, scalable, and adaptive measurement.

‍

Falen Milton is VP of People at Workera, where she leads people and talent. She built her career in talent acquisition before joining Workera four and a half years ago, and now works at the intersection of recruiting practice and AI adoption, navigating the same shifts her peers across the industry are facing in real time.

The Four Signals Every Hiring Process Runs On

Most hiring processes draw on four broad buckets of evidence. First, some form of resume or CV, often carrying education history, professional lineage, and a narrative of how the candidate arrived where they are. Second, pedigree, meaning the schools and employers attached to that history. Third, a referral, or the sense that someone in your network knows someone who knows this person. Fourth, at least one interview, conducted by a human or increasingly by AI. Interviews have the longest track record and the most science behind them, but that science only pays off when the process is calibrated so responses are comparable across candidates and comparable over time. Most companies never take that step. What is left is gut feel, frequently labeled culture fit, which in many cases means: do they look like me, talk like me, and share my background? Humans gravitate toward the familiar, and that pattern matching quietly produces a workforce with far less variety in it than anyone intended.

AI Did Not Break These Signals. It Exposed Them.

Every one of the four signals carried noise long before generative tools arrived. AI ballooned the flaws. A resume was always partly a test of how well someone writes and frames their own impact, and whether they knew which keywords to include. Now the writing is outsourced. What a candidate chose to put on the page, and just as tellingly what they chose to leave off, used to be a genuine read on judgment. That discernment is gone when a tool supplies the template and parses the content in. Volume compounds the problem: roughly eleven thousand applications are now submitted every minute, because a candidate who once fielded a single all-purpose resume can generate fifty tailored versions in seconds. Ninety-one percent of managers report encountering or suspecting AI-generated interview answers. As Sullivan put it, the hiring process has started to look like one agent talking to another, and the hard part is finding the human in the exchange.

Using AI Is Not the Problem. Slop Is.

Neither panelist wants candidates to stop using AI. Milton actively wants to see it, because these are the tools people will reach for on the job. Her gut check is a set of practical questions: is this a tool the candidate will actually have access to once hired, did they quality-check what came out of it, are there still typos, does it make sense, does it read human? What she is screening for is not abstinence but authorship. Sullivan framed the stakes plainly: if you are getting slop during the interview, you will get slop on the job. That makes AI fluency a legitimate hiring signal rather than a threat to one. The caution both raised is on the questions themselves. Asking candidates something open enough to show their own style and thinking works well. Asking how many golf balls fit in a swimming pool does not, because a signal that is not job related is not a valid signal, and it introduces fairness exposure at the same time.

The Catch-22: Why One More Step Backfires

Fifty-nine percent of hiring teams suspect misrepresentation. Only nineteen percent feel confident they could confirm it. That gap creates an obvious temptation: add another check, another verification, another loop. Milton's argument is that the extra hurdle filters for exactly the wrong person. Anyone genuinely misrepresenting themselves has their material saved in a bot and clears the additional step without friction. The candidate who is employed, honest, and short on time is the one who drops out. Sullivan added two costs on top of that. Every added stage extends the hiring cycle, which means a business-critical role stays open longer and the eventual start date slides. And if extra steps get applied to some candidates and not others, the experience stops being uniform, which is where legal defensibility starts to erode.

Name the Suspicion Instead of Working Around It

The better move when something feels off is to say what it is. Unnamed discomfort does not disappear; it resurfaces as vague language about ways of working or fit, and at that point it reads as bias, because nobody can articulate what actually happened. Milton's approach is to identify the specific claim in question and talk it through directly. Sullivan described how the team pressure-tests that instinct internally, walking back through the exact moment in the interview or the exact line on the resume that triggered the reaction, then turning it into a real question for the candidate: you are applying at this level, tell me about a time you did work at this level. She also offered a reframe worth sitting with. Assume everyone is managing impressions, because that is what a hiring process is. Some people are simply better at it, and that skill is often job relevant. The line that matters is not whether someone is presenting their best self. It is whether they are presenting a capability they do not have.

What a Mis-Hire Actually Costs

The bill arrives in layers. Recruiter time and spend. The revenue or output lost while the seat sat empty. The full investment in ramping someone who was never going to work out. Then, when it becomes clear the hire has failed, an exit is rarely quick or clean: performance has to be documented, coaching has to be offered and recorded, and a paper trail has to exist before anything else can happen. Underneath the accounting is a morale cost that Milton flagged as the one people underestimate. A team that has been waiting for relief gets someone who cannot deliver it, then watches that person leave. Trust in the process takes the hit. And there is a cost on the other side of the table too. Honest candidates spend real hours researching the company, preparing, and taking time away from their current role, which is part of why chasing them through additional hoops is more expensive than it looks.

Internal Mobility Is the Underused Answer

Asked whether any of this applies inside the organization, Milton's response was that internal mobility is where the opportunity is most obvious and most neglected. Hiring externally is expensive, and in many cases the capability you need is already on payroll. Someone internal comes with months of performance history, an established relationship, a known narrative, and skills that have been demonstrated rather than described. Trialing that person in a new function is dramatically cheaper than sourcing from outside, and if it does not work, you still have the data and the relationship. What makes it hard today is retrieval. Her point to recruiters was direct: a searchable database of verified, job-relevant skills across your existing workforce would be recruitment gold, and that is the direction things are heading.

The Skills Passport

Sullivan's five-year view starts from what AI made cheap. Assessment was historically expensive to build and expensive to maintain, and anything bought off the shelf rarely fit the exact need, which is why rigorous measurement stayed out of reach for most organizations. That constraint is gone. Job-relevant signal can now be generated quickly and at quality. Her prediction is that measurement stops being an event and becomes a record. Children are already assessed continuously through school, and universities assess further; the stream simply stops at the moment someone enters the workforce, precisely when employers need it most. She calls the alternative a skills passport: a portable, verified account of what a person can do and, more importantly, how that has changed over time. With the half-life of skills shortening, the pattern of how someone has reskilled themselves for new opportunities tells you more than any single snapshot. She thinks five years is the conservative estimate and two is plausible. Milton compared the transition to the arrival of the credit card. Nobody argues now that a chip and PIN made buying groceries less human; it made the transaction possible and verifiable. Skills verification will feel intrusive for the first year or so, and then it will feel like a background check: expected, unremarkable, and part of the process.

Is Proctoring the Answer? No.

Sullivan's answer was immediate. Proctoring is one of the fastest ways to put off the right candidate, and not because that candidate wants to cheat. Being watched is uncomfortable, and it broadcasts a lack of trust before the relationship has started. It can still be a useful tool when the design is thoughtful, which is the standard Workera set for its own recently released proctoring: minimally invasive for the candidate, rich behavioral data for the hiring manager, and no video or audio monitoring, on the view that an AI platform has to earn trust with the people using it first. Her broader position is that cheaters will cheat regardless of the controls in place. The realistic goal is deterrence and friction against the most common methods, not elimination. The more durable fix is the one both panelists kept returning to: rather than bolting another stage onto the process, build the verification into the live interview you are already running.

‍

‍