Candidates are job hunting with AI. Most employers aren't ready for it.

More job seekers are starting to explore opportunities through AI tools like ChatGPT. According to iHire's 2026 State of Online Recruiting, 11.1% of job seekers now use AI platforms in their search, roughly one in nine. Someone might ask ChatGPT for "remote marketing jobs in healthcare paying over $120,000" and receive a summary of relevant opportunities with links to learn more or apply. They may never open a traditional job board or visit your career site's home page.

I think this is the biggest change in recruiting technology since job search moved to phones, and most of our industry hasn't caught up yet. When we talk about AI in recruiting, we still mostly mean making recruiters faster. That's real. LinkedIn's 2025 Future of Recruiting research found that recruiters using generative AI saved about 20%of their work week, which is a full day back every week. But the question I'd ask first is a simpler one. Can candidates who search with AI find you at all? 

Can an AI assistant actually read your jobs?

An AI assistant can only recommend a job it can read. That sounds obvious, but you'd be surprised how much job content is hard for these tools to use. Pay is missing. Titles use internal names like "Associate II, Ops" that no candidate would ever type. Some postings sit on pages a crawler can't even load. For more on making your jobs visible to AI search, see our post on recruitment SEO and AEO.

A recruiter reading that posting can work around the gaps. An AI assistant comparing fifty jobs for a candidate mostly won't. It will skip yours and recommend the one with clear pay, a plain title and a complete description.

And this is just the start. Agent tools like Muse are beginning to search for jobs on a candidate's behalf, and I expect they'll be applying for them before long. Once software does the first pass of a job hunt, the employers who show up will be the ones whose jobs are complete, structured and easy to reach.

Candidate experience is where I'd push harder

For years, the standard advice on candidate experience has been pretty cautious: fix the basics, remove unnecessary steps, and be careful not to over-automate. That's still good advice, but I'd go further now. Candidate experience is where I think most employers should be adopting technology more aggressively, because people who find jobs through AI expect quick, specific answers, and a lot of application processes still can't give them any.

Monster's 2026 Application Black Box Report found that 60% of surveyed U.S. job seekers said their biggest frustration was not knowing whether a human had reviewed their résumé. That's more than half of them. What bothers me is that the ATS almost always knows where an application stands; the candidate just can't see it. An AI assistant that can answer "where is my application, and what happens next?" closes that gap, and it doesn't take any recruiter time.

The same goes for the application itself. If someone found your job through a conversation with ChatGPT, it's jarring to then make them create an account and retype their work history into a long form. Conversational applications, résumé parsing that actually works, and plain-language answers to screening questions are all things I'd expect most employers to be testing within the next year. Some of what's possible here is really neat.

Where people, and the law, still make the call

There's a real tension here, though. In her July post on AI in recruiting and the case for a human-centered approach, NAS COO, Jennifer Henley pointed to a 2025 Gartner survey that found only 26% of job candidates trust AI to evaluate them fairly, so roughly one in four. From the candidate's side, that raises an interesting distinction. The same people who are happy to let ChatGPT find their next job don't necessarily want it deciding whether they get an interview.

I think that's a fair position, and employers should respect it. Use AI to help candidates find you, apply and stay informed. Keep people on the hiring decisions, and make it easy for a candidate to reach a person when their situation doesn't fit the process.

Compliance is the other reason to be careful here, and I don't think it gets enough attention. Rules on AI in hiring are arriving faster than most recruiting tools have adapted to them. New York City requires an independent bias audit, and notice to candidates, before an employer uses an automated tool to screen or rank them. Illinois requires consent before AI analyzes a video interview, and Colorado has passed a broader law covering AI used in hiring decisions. My impression is that many tools on the market haven't really considered AI compliance, and when something goes wrong, the liability usually lands on the employer that used the tool.

Jumping into more AI in recruiting without thinking through compliance can get you into deep water quickly. Before you turn a tool on, I'd ask the vendor who audited it, what candidates are told about it, and how a candidate can ask for a person instead. If the vendor can't answer those questions, it's a good idea to partner with someone who can help you navigate the compliance side and make sure what you're delivering is compliant. That's part of how we approach any solution at NAS. What we put in front of candidates for our clients has to hold up when a regulator, or a candidate, asks how it works.

How do you know any of this is working?

None of this is worth much if you can't tell whether it's working. SHRM's 2025 benchmarking research found that only 20% of organizations track quality of hire. That's one in five. In my experience, even fewer can connect where a candidate first found a job to whether that person was eventually hired.

Honestly, that connection is the part of recruitment marketing I find most interesting. At NAS, we run the top of the hiring funnel for our clients. We get their jobs in front of the right candidates wherever those candidates are searching, and through ACTIVATE Analytics, we tie that spend to the applications and hires it produces in the client's ATS. As we explain in Does ACTIVATE Work With Your ATS?, this connection doesn't require replacing the ATS you already use. I think a team that specializes in that work, connected to the ATS you already use, will see a lot more of the funnel than the advertising module that comes bundled with an all-in-one suite. Matt Adam makes a similar point in Recruiting Is No Longer Just About Filling Jobs.

AI search adds a new line to that report. If you don't know how many of your applicants are arriving through ChatGPT or another AI assistant, I'd start tracking it now, while the number is still small. It's also worth setting a baseline for application conversion and cost per qualified applicant at the same time, so you can tell whether changes to your job content or application process actually made a difference.

Here's a quick experiment

If you want a quick sense of where you stand, try this. Open ChatGPT or Claude and ask it to find jobs at your organization: a specific role, in a specific city, at the pay you actually offer. Then ask it what it knows about working there.

If your jobs don't come back, or they come back with the wrong pay and an old description, that's what a growing share of your candidates are seeing. I think it's a better place to start your next technology decision than any vendor's feature list, including ours.