You've probably seen it: the veteran recruiter with decades of real, hard-won expertise, suddenly posting deepfake-smooth recruiting videos and a headshot that looks like it was run through a filter labeled "aggressively confident." The hardest part to understand? Someone who's actually great at this job is choosing to look like she just discovered the internet.
That’s just one end of the trap. The other end are the recruiters and hiring teams who've made AI feel like contraband, where using it at all gets treated as a shortcut around real thinking.
Neither one is a strategy. Both are just "whether we use AI" standing in for an entire personality, when the actual skill nobody's talking about is knowing how to use it well.
The reality is that the best hiring actually happens somewhere between "we don't touch AI, full stop" and "AI touches everything, always." Most growing companies are stuck at one end or the other, and both spots are costing them stronger hires.
We've already covered how to evaluate a candidate's AI literacy in an interview. We’re flipping the coin here and asking the same question of the hiring manager.
Refusing AI in hiring doesn't make a process more human. It usually just makes it slower, and it puts hiring managers at a real disadvantage against candidates who are already fluent in the tools.
The imbalance is already showing up in the data. Job postings are drawing volumes that human-only screening was never built to handle, and candidates have gotten very good at using AI to look qualified on paper, whether or not the underlying skill is there.
Harvard Business Review's recent research, based on interviews with 120 talent-acquisition leaders and an analysis of over 6,000 recorded screening calls, found that traditional hiring signals like a polished resume or a smooth recorded interview answer no longer reliably measure competence. They increasingly measure how well someone can produce a good artifact, with or without AI's help.
Companies that treat any AI use as a violation of "how we do things" aren't protecting quality but instead opting out of the tools that could actually help them spot that gap.
The overcorrection isn't better. Handing AI the parts of hiring that require judgment doesn't remove bias or inefficiency. It just moves the decision-making somewhere harder to see and even harder to walk back.
The same HBR research makes the underlying problem explicit: Remote, asynchronous, low-accountability formats are exactly what AI is best at gaming.
When every job description, every recruiter message, and every interview question is AI-generated, a company isn't running a smarter process. It's running a process candidates can now predict and prep for, which quietly strips out the signals that actually mattered in the first place, like how someone thinks under pressure, what they ask, and whether their voice sounds like an actual person.
And full automation doesn't guarantee better outcomes on its own. Leadership IQ's long-running research on new-hire failure, based on more than 20,000 hires and thousands of hiring managers, found that 89% of hiring failures come down to attitude, coachability, and fit, not a lack of technical skill.
Technical skill is exactly what AI screening is best at verifying, and exactly the smaller slice of what actually determines whether a hire works out.
So the fix here isn't removing human judgment and taking your brain out of the picture. It's making sure AI is aimed at the parts of the process that actually benefit from speed and structure, not the parts that require a read on a person.
The companies getting this right treat AI the way good agentic tools are already built to work in HR: It collects information, flags what's missing, drafts a first pass, and organizes the workflow, but a person still makes the call. That framework translates directly to hiring.
You can start with an AI draft of the job description. But, the most important part is to edit it until it sounds like your company instead of every other AI-generated posting in the applicant's inbox. If a candidate can tell your outreach was copy-pasted from a prompt, you've already lost some trust before the first call. (More on making AI-assisted hiring copy actually sound human here.)
AI is genuinely useful for building out a structured question bank, organizing notes across candidates, and flagging areas worth probing further. What it’s not useful for is deciding who felt right in the room. That’s a judgment call, and it’s your very human job.
For growing or creative teams fielding hundreds of applications on lean recruiting bandwidth, AI-assisted screening is a legitimate time-saver. According to SHRM's latest talent trends research, 51% of organizations already use AI somewhere in recruiting, most commonly for writing job descriptions and screening resumes. Use it there. Don't let it make the final call on who gets an offer.
HBR's fix for the broken-signal problem is to shift early-stage hiring toward live problem-solving, work samples, and real-time reasoning, the parts of the process that are hardest to fake or outsource.
And that's a good rule of thumb for any creative or startup team; the harder something is to fake with AI, the more weight it should carry in your decision.
Every hiring manager reading this probably thinks their own process already lands in that middle ground, but candidates don't always experience it that way.
Here's what the two extremes actually look like from their side of the table.
Picture a UX Content Designer three rounds deep into two different interview processes at once.
Company A has a strict no-AI policy, and it shows up everywhere.
The application requires her to attest that no AI tooling touched any portfolio piece, full stop, even though she used it the way most working writers do now: to test five microcopy variants before picking the one she actually shaped.
When she mentions this in the interview, the tone shifts. It doesn't read as honesty to them. It reads as a red flag. She leaves the process feeling like her actual craft, the judgment calls about tone, hierarchy, and voice, was overshadowed by a policy that couldn't tell the difference between using a tool and outsourcing the thinking.
Company B goes the other way. Her first two rounds are fully automated with a chatbot screen, then an AI-scored writing exercise, with zero human contact until the final round.
The interview questions feel generic, clearly pulled from a template rather than her actual portfolio.
When she asks a pointed question about how the team actually collaborates with product and eng, the interviewer can't answer; that part of the process was never built for two-way conversation. She gets an offer, but she has no real read on the team, and neither does the company, because nobody on the human side ever had a real conversation with her.
Same candidate, same skill level, two ways to lose her.
Company A filters out a strong hire over a false signal. Company B extends an offer without anyone actually vetting the fit. Neither company will know why she said no, or worse, why the hire didn't work out.
Not every stage of hiring needs an AI touchpoint to prove a company is forward-thinking. Sometimes the more sophisticated move is knowing where to leave it out entirely, especially in the moments where chemistry, curiosity, and communication style are what you're actually evaluating. Those signals get lost fast when too much of the process runs through a model.
This is also where gauging comfort matters. Not every candidate, client, or colleague has the same expectations around AI use, and reading that room is its own form of AI literacy, separate from knowing how to use the tools well.
Most hiring managers don't think of themselves as extreme. Answer honestly and see where the process actually lands.
Mostly "yes" on 1 and 4: You're closer to the no-AI camp than you think, and it's probably costing you time and candidates who'd otherwise be strong fits.
Mostly "yes" on 2 and 6: You've handed off more judgment than you realize. Worth pulling the decision-making back to a person before the next req opens.
Mostly "yes" on 3, 5, and 7: You're in the middle ground this piece is arguing for. Keep going.
AI proficiency in hiring was never about using AI as much as possible. It's about having the judgment to know what to hand off and what has to stay distinctly human.
If you get that balance right, the payoff shows up on the other side of the offer letter, too, with candidates who have been tested on real skill and fit starting their jobs with a clear sense of what it actually entails and confidence in the value they bring to the team.
That's the piece that gets missed when companies plant a flag on either extreme.
The goal was never "more AI" or "less AI." It's a hiring process good enough that it survives contact with either one, and a team that already knows what real AI reskilling looks like for creative talent once they're in the door.
None of this is easy to figure out alone, and it shouldn't have to be. Sorting out which parts of your hiring process deserve AI and which parts need an actual human read is exactly the kind of thing that's clearer with a second opinion in the room. That's what Artisan Talent is here for, to be a sounding board that's already sifted through the AI noise, so the judgment calls left in your hiring process are the ones that actually matter. Talk to the Artisan team about building a hiring process that holds up on both ends.