The data is clear: AI adoption in job search has moved from novelty to norm. Multiple surveys across 2025 and early 2026 put the figure at roughly three in four job seekers using AI tools in some part of their search. ChatGPT, Claude, Gemini, Copilot, and a growing number of purpose-built platforms.
The tools are good. The way most people use them is not.
I see the output every day. CVs that use the same three phrases in the professional summary. Cover letters that hit every keyword but carry no voice. LinkedIn summaries that read like they were generated from a template, because they were. Application materials that are technically competent and strategically invisible.
The problem is not AI. The problem is that most people use AI for the thing it is worst at (positioning) and ignore the things it is best at (research, preparation, and analysis).
Where AI is genuinely excellent
I want to start with where the tools add real value, because the answer is "more places than most people realise." They are just not the places most people start.
Company research
This is the single best use of AI in a job search. Ask Claude to give you a briefing on a target company: what they do, their recent funding history, leadership changes, competitive position, and the challenges a business at their stage is likely facing. You will get in ten minutes what would have taken two hours of manual research.
That research is the foundation of a strong direct approach. It is the difference between a generic application and a tailored message that demonstrates you understand the business. I wrote about why this matters in how to approach companies directly.
Market mapping
Use AI to identify companies in a specific sector, at a specific stage, with a specific profile. "Show me PE-backed UK businesses in consumer or industrial sectors with revenues between £30m and £200m that have made acquisitions in the last 18 months." That query, run through Claude or ChatGPT with web access, gives you the beginning of a target list.
The output needs verification. AI tools hallucinate, particularly on specific company data. But as a starting point for a target list that you then verify manually, it saves hours.
Interview preparation
Ask AI to research the person who will interview you. What have they posted on LinkedIn recently? What is their professional background? What are the likely priorities of someone in their role at a company at this stage?
Then use AI to anticipate the questions. Not generic interview questions from a blog post. Specific questions that this interviewer, in this role, at this company, in this market, is likely to ask. "What questions would a CFO of a PE-backed consumer business ask a candidate for Head of Finance, given that the business is six months post-acquisition and integrating two finance teams?"
The quality of the output depends entirely on the quality of the prompt. Give AI the context, and it gives you preparation that is genuinely useful.
Salary benchmarking
AI tools can aggregate data from multiple sources to give you a reasonable range for a specific role in a specific sector and location. The data is imperfect, but it is better than walking into a negotiation with no reference point.
Where AI makes you invisible
Now for the part that matters more.
CV writing
This is where most people start, and it is where the damage is done.
When you ask AI to "improve my CV," it applies a statistical model of what CVs generally look like. It tightens the language. It adds action verbs. It restructures bullets into the accomplishment format. The output is grammatically better and strategically identical to the output it produces for everyone else.
I wrote about this in detail in AI CV builders vs human CV review. The core problem is that AI cannot position you because it has no model of how a specific hiring manager reads a specific CV for a specific type of role. It optimises language. It does not optimise strategy.
The convergence effect is real. When I read CVs for a search, the AI-written ones are increasingly recognisable. Not because the writing is bad, but because the writing is the same. The same structures, the same phrases, the same generic value propositions. In a shortlist of eight candidates, the ones that sound like templates get less attention, not more.
Cover letters
AI-generated cover letters are even more recognisable than AI-generated CVs, because the format is more constrained. "I was excited to see your posting for [role] at [company]. With my [X years] of experience in [field], I am confident I can contribute to your team's success."
Every recruiter in the country has read that paragraph a thousand times. It is the written equivalent of a blank stare.
LinkedIn summaries
The same convergence problem applies. AI-generated LinkedIn summaries tend to follow the same arc: who I am, what I have done, what I believe in, what I am looking for. The structure is fine. The voice is absent. When a recruiter searches LinkedIn and scans five profiles, the one that sounds like a person stands out. The four that sound like ChatGPT blur together.
Anything that a hiring manager or recruiter reads to assess who you are. If it sounds like it was generated, it sounds like everyone else. And sounding like everyone else is the definition of poor positioning.
The right relationship with AI
The pattern is simple. Use AI for tasks where speed and breadth matter and where the output is a starting point you will refine: research, mapping, preparation, analysis. Do not use AI for tasks where your voice and positioning need to come through: the CV, the cover letter, the LinkedIn summary, the direct approach email.
Think of AI as a research analyst, not a ghostwriter. You would not hand your CV to a junior analyst and say "write this for me." You would ask them to research the target company, pull the data, prepare the briefing, and then write the document yourself based on what they found. That is the right model for AI in a job search.
If you do use AI for drafting, treat the output as a starting point, not a finished product. Run the draft through your own voice. Remove anything that sounds like it could have been written for anyone. Add the specific details, examples, and judgements that only you can provide. If the final version sounds like it could have come from ChatGPT, it needs more work.
The human layer
The thing AI cannot do is tell you how your materials read to the people who make hiring decisions. It can improve the language. It cannot tell you that your career reads as a generalist when you need it to read as a specialist. It cannot tell you that your professional summary leads with the wrong thing for the roles you are targeting. It cannot tell you what a hiring manager will think when they compare your CV against the other seven on the shortlist.
That assessment requires the experience of having been on the other side of the table. Having read thousands of CVs against real mandates. Having watched shortlists get built and cut. Having heard hiring managers explain exactly why they said yes to one candidate and no to another.
The CV Intelligence Report provides that human layer. Every section of your CV assessed from a recruiter's perspective, with the reasoning annotated. Two tailored CVs. A LinkedIn assessment. And the positioning strategy that no AI tool can generate, because positioning is a judgement, not a language task.
Use AI for what it does well. Use a recruiter's perspective for what it cannot.
Share
LinkedIn
Michael Muir
Founder · The Other Side
Twenty years placing candidates across high-calibre boutiques through to FTSE 100 companies. Thousands of CVs a year. Writes “Notes from the Desk” on how hiring decisions actually get made.
Read more about Michael
