A good resume used to show, at least partially, whether someone could structure experience and identify what mattered. Today AI can do that in minutes.
The same is true for cover letters, interview preparation, standard answers, and take-home assignments. The problem is not that candidates use AI. The problem is that companies still evaluate signals that have become much easier to manufacture.
A polished resume is weaker evidence now
Clear language, strong verbs, convincing achievements, and perfect structure can help a candidate communicate, but they say little about how deeply that person understands the experience described.
I would treat the resume mainly as a map for the conversation. It tells you what episodes to explore, but it should not carry too much weight by itself.
Test the depth of the story
If someone says they improved a process, built a system, or shortened delivery time, do not stop at the result.
Ask what they personally did, what alternatives existed, who disagreed, what failed on the first attempt, and what they would do differently today.
People with real experience usually remember the rough edges. A story that sounds perfect often becomes much weaker after a few concrete follow-ups.
Standard interview questions are even less useful
“What are your weaknesses?”, “Tell me about a conflict”, “Where do you see yourself in five years?” — almost perfect answers can be prepared in advance.
And this is not only about AI. We have spent years teaching candidates how to pass interviews correctly.
The more predictable the process is, the more it measures preparation for the process rather than ability to do the work.
Change the constraints during the conversation
Take a realistic case and, after the first answer, change one constraint: the budget is cut in half, the deadline moves forward, a key person leaves, or the business changes priority.
The important part is not whether the new answer is perfect. It is how the candidate restructures the reasoning.
AI can prepare someone well for a known question. It is much harder to prepare in advance for every change in context.
Do not ban tools people will use at work
If employees are allowed to use AI after they are hired, it is strange to design the entire interview around proving they never use it.
It is more useful to separate two things: ability to use the tool and possession of real expertise.
You can allow AI for part of an exercise, then ask the candidate to explain the result, find weaknesses, modify the solution, and defend the tradeoffs.
Watch the quality of the candidate's questions
Strong specialists rarely just answer. They clarify context, notice contradictions, and try to understand the real problem behind the request.
That signal becomes more valuable in a world where answers are getting easier to produce while good questions remain difficult.
Hiring becomes less convenient, but more honest
AI did not make all candidates the same. It made many external signs of preparation look the same.
A good hiring process should therefore evaluate packaging less and focus more on how a person thinks, explains past decisions, responds to new context, and takes responsibility for outcomes.
If a company still hires mainly on perfect resumes and perfect answers, the problem is no longer AI. It is the hiring process.