Workology Podcast

The three-phase AI maturity model (automation → signal extraction → skill validation) and the critique of the requisition-based model offer some practical framing, but the episode heavily repeats the same thesis (resumes are bad) multiple times and leans on broad, established IO-psychology consensus rather than surfacing new ideas per minute. There is meaningful padding and throat-clearing throughout.

“it's moving away from what candidates, um, are telling you about themselves to them actually showing you something about themselves. So it's really a move from tell me to show me”

“The first phase, kind of the entry level adoption is going to be bringing AI for automation... phase two... signal extraction... The ultimate is when you're using artificial intelligence to help you with skill validation”

The sharpest and most genuinely contrarian point - that the job requisition model itself should be abolished in favour of a standing skills-profile approach - is interesting and underexplored in mainstream HR discourse. Everything else (resumes lack predictive validity, skills over pedigree, garbage-in-garbage-out on AI data) is standard fare that circulates widely in HR tech media.

“another foundational element of hiring was the job requisition. And so that's another thing that I wish would just die is the job requisition”

“The processes we're replacing are broken and full of bias to begin with”

Mike Hudy is a legitimate industrial-organisational psychologist with genuine practitioner credentials - CSO at HireVue, previously CSO at Modern Hire and founding EVP of Science at Shaker International - giving him real standing to discuss predictive validity and assessment science. However, much of the episode is spent in vendor-advocate mode promoting HireVue's product suite, which mutes the independent intellectual value.

“I'm an industrial organizational psychologist, and in our field, we've done actually more than a century of research on what hiring tools are actually most predictive of success on the job”

“We've been doing assessments, uh, and interviews for over 20 years. Uh, and so we have lots of data where we directly measure skills of candidates”

The episode offers some concrete illustrative numbers (12 days to 3, 30/60/90-day retention windows, 6 - 12 months to proficiency) and names specific post-hire metrics (NPS, sales conversion rates, supervisor rehire surveys), but cites no actual research findings, no named customer examples, no percentages from HireVue's own dataset, and no referenced studies despite repeatedly invoking 'a century of research.'

“Before I brought AI to my process, it was taking me 12 days to hire somebody. Now I'm down to three”

“If we have one opening and 10,000 people apply to that opening, all 10,000 are going to be seen, all 10,000 are going to be evaluated the same way”

The host asks predictable, sequentially logical setup questions and lands one solid ROI-timeline follow-up, but never challenges any of HireVue's vendor claims, never probes the limitations of AI-scored interviews, and allows repeated self-promotional assertions to pass unchallenged. Several questions are essentially rephrases of the prior answer.

“I feel like we have been talking about the, a resume being ineffective for a really long time. And I'm excited that we have technology and tools that can help assist with that”

“So what I hear you saying is it's really not a short term ROI. It sounds like a year, 18 months, 24 months”