Skip to content
Back to writing

Ergonomics and AI: what the tool displaces in practice

Opening note. What AI concretely displaces in the ergonomist's practice and in organizational work — and the analysis space it opens.


Ergonomics and AI: what the tool displaces in practice

I am reopening this site to write, publicly and regularly, what I really think about the profession. The first thing I need to say is about artificial intelligence. So I may as well start there.

The short position

AI is not a threat to ergonomics. It is a logical continuation of its project. Ergonomics has always tried to adapt the tool to the human, to increase performance without degrading the person. Language models, transcription tools, specialized assistants are exactly that: tools that, when used well, augment the human. Used badly, they crush the human. Like every technology before them.

For an ergonomist, AI is now an accessible and low-cost quality lever. It does not do the work in their place — it does not — but when used well, it frees time for field work and analysis. This is a concrete opportunity for the practice, provided we keep the steering.

What it changes concretely for the ergonomist

My field feedback after months of integration:

Collection. Transcribing an interview no longer consumes days. A few minutes are enough to get a clean, timestamped, searchable transcript. I reread better because I read more.

Analysis. Crossing hours of observation with HR documents and a DUERP to surface patterns is a task where AI, properly steered, accelerates what I would do alone. Provided I keep the steering.

Restitution. Writing a report that is both rigorous and readable by an executive committee is a style exercise. Writing assistance does not replace judgment, but it reduces the distance between thought and page.

Client relationship. I am more present and less buried in paperwork. Time saved on document production is reinvested in field time and discussion. That is net value for the client.

What it changes for clients

This is the other half of the subject, and the one people discuss much less.

Operators, managers, support functions — everyone is already using AI, with or without authorization, with or without a frame. Cases come up from the field:

  • service schedules drafted by ChatGPT without anyone checking recovery time;
  • follow-up emails to users written by an assistant, with phrasing that creates more conflict than it resolves;
  • “AI-assisted” psychosocial-risk analyses producing generic reports, archived and never reread;
  • and the reverse: operators saving real time every day through good use of an assistant, with no procedure formalizing it.

There is, in all this, a field of cognitive ergonomics that is massively understudied. What new mental loads appear when people work with an assistant? What new rule conflicts emerge when a machine suggests a solution the operator knows is wrong, but the manager validates it because it was “generated”? What new safe practices need to be formalized?

The ergonomist has something to say about these questions. Few people have our position: observing real work, not prescribed work, not imagined work. It would be a shame to leave this ground to IT consultants who will never watch a workstation for two hours.

What I refuse

I refuse the posture of the ergonomist who stays away from AI on principle, as if refusal carried some kind of nobility. It is not nobility. It is a lack of professional curiosity, and sometimes a little fear, which is human but should not become a public position.

I also refuse the opposite posture: the promise seller deploying tools without understanding what they do to the activity. This is exactly the kind of transformation that, when analyzed three years later, has created more psychosocial risk than it solved.

Between the two, there is a narrow and useful space: an ergonomist who uses AI with discernment, observes what it does to work, and writes what he sees. That is what I intend to do here.

What this site becomes

I am opening a space for notes, alongside the framing pages. It will include field observations, method points, anonymized cases, and reflections on AI at work. The engagement pages remain; the notes extend them.

To reach me: email or the Julien Talbot home page. To follow the writing: the RSS feed.

The rest will come at the pace where I have something to say that deserves to be published.