Abstract: AI hasn’t lower anybody’s workload, it’s simply made folks produce extra, quicker. The issue is no person’s modified what occurs after: work nonetheless will get caught in the identical approvals, the identical conferences, the identical supervisor’s inbox. Until organisations rethink these processes, all that saved time is misplaced.
The dialog round AI at work remains to be largely centered on instruments:
- Which platform ought to we purchase?
- How shortly can we roll it out?
- How will we encourage extra folks to make use of it?
- And the way will we construct sufficient AI functionality throughout the organisation?
These are essential questions however, following various management activation occasions I’ve been operating with Occupational Psychologist, Dexter Winters, I’ve been doing quite a lot of occupied with the problem coming subsequent.
Now we have witnessed time after time that after folks begin utilizing AI efficiently, one thing moderately uncomfortable occurs: they produce extra work.
Extra: stories, shows, evaluation, concepts, content material, suggestions.
All created at a pace that may have been not possible only some years in the past. On the floor, this seems like productiveness however producing extra doesn’t essentially imply turning into extra productive.
The truth is, many individuals I converse to are telling me that they really feel busier, extra overwhelmed and below better strain than earlier than.
AI has elevated the move of labor, however most organisations haven’t modified the system round it.
The fireplace hydrant and the O-ring
That is the place the plumbing analogy is available in (my thanks goes to Dexter for the inspiration on this one!).
AI has turned on a hearth hydrant of labor, however we’re nonetheless making an attempt to push the whole lot by the identical outdated pipework.
Someplace in that system is a tiny O-ring which is often a supervisor, senior chief or management staff who is anticipated to overview, problem, approve or log off the whole lot now arriving at twice the pace.
The worker might need used AI to scale back three hours of labor to 45 minutes, however the output nonetheless enters the identical approval course of. It waits for a similar supervisor, will get mentioned in the identical assembly and ultimately joins the identical decision-making queue.
The creation of the work has accelerated. The organisation has not. Ultimately, one thing should give.
The supervisor turns into overwhelmed, requirements start to slide or folks cease reviewing work as rigorously as they need to. Alternatively, the organisation creates an infinite quantity of completely acceptable work that no person notably wants.
This is the reason we have to cease treating AI adoption as purely a expertise or expertise problem. It’s quickly turning into an organisational design problem.
Many individuals I converse to are telling me that they really feel busier, extra overwhelmed and below better strain than earlier than
Extra output isn’t the identical as extra worth
Particular person productiveness and organisational productiveness aren’t the identical factor.
If I save an hour writing a report, that’s a person productiveness achieve. But when the report nonetheless spends 4 days ready to be accredited, requires three conferences and is then rewritten as a result of no person agreed its function within the first place, the organisation has gained little or no.
That is the place quite a lot of anticipated AI return on funding may quietly disappear.
Organisations could also be measuring what number of staff are utilizing an AI instrument; what number of licences have been activated or how a lot time folks say they’ve saved. These measures might be helpful, however they don’t inform us whether or not work is shifting by the organisation extra successfully or creating higher outcomes.
The extra beneficial questions are:
- What occurred to the time saved?
- Did selections enhance and, if that’s the case, why and the way? If not, why not?
- Did prospects obtain a greater service? How do we all know?
- Did groups cut back low worth exercise? How do we all know?
- Had been folks in a position to concentrate on work requiring judgement, creativity or human connection? If not, what bought in the best way?
If we can’t reply these questions, we could merely be utilizing AI to provide extra widgets.
We’d like new norms for AI-enabled work
There’s one other subject that many organisations haven’t but addressed: individuals are being inspired to make use of AI and not using a shared understanding of what good AI-enabled work seems like.
Extra inquiries to ask which can assist this journey embody;
- When ought to AI be used, and when ought to it not?
- What degree of human overview is anticipated?
- Who owns the ultimate resolution?
- Does each AI assisted doc want approval?
- When can staff act autonomously?
- What proof ought to accompany a suggestion?
- How do folks declare the place AI has contributed?
Notice: These questions can’t be answered by an AI coverage alone. Insurance policies set boundaries, however folks additionally want sensible working norms that replicate what actually occurs inside their roles and groups.
With out these norms, managers grow to be the security web for the whole lot. Workers are advised to experiment however nonetheless really feel the necessity to search reassurance earlier than appearing. Leaders need innovation whereas sustaining the identical controls, hierarchies and approval processes that existed earlier than AI. That isn’t transformation. It’s outdated work shifting quicker.
Particular person productiveness and organisational productiveness aren’t the identical factor
L&D has a a lot larger position to play
That is the place studying and improvement must broaden its contribution.
AI coaching can’t cease at instructing folks how you can use a instrument or write a greater immediate.
Workers want to grasp how you can choose AI outputs, the place human experience provides worth and how you can collaborate successfully with AI. Managers want help to revamp work, set clearer expectations and lead groups whose capability could change considerably.
This additionally means serving to groups study their work truthfully with questions like:
- Which actions nonetheless have to exist?
- Which approvals genuinely cut back danger?
- Which conferences at the moment are redundant?
- The place are selections getting caught?
- What ought to folks do with the time AI releases?
If we construct functionality with out addressing these questions, we danger creating quicker staff inside gradual organisations.
The subsequent part of AI adoption
The subsequent part of AI adoption might be much less about entry to expertise and extra about work cadence, resolution making and organisational design.
It should require leaders to clarify why AI is a part of the organisation’s future, what which means for workers and, equally importantly, what it doesn’t imply.
It should require managers who really feel assured sufficient to steer otherwise moderately than merely dealing with extra. And it’ll require organisations to determine the place the AI Dividend™ (the time, power and capability launched by AI) ought to go.
The fireplace hydrant has already been switched on. The query now’s whether or not organisations will redesign the pipework or await the strain to reveal its weakest factors.
The subsequent part of AI adoption might be much less about entry to expertise and extra about work cadence, resolution making and organisational design
Actionable insights
- Ask what occurred to the hours AI saved final month. If no person can reply, they in all probability didn’t go anyplace helpful.
- A 3-hour activity achieved in 45 minutes nonetheless hits the identical four-day sign-off. Repair the sign-off, not simply the duty.
- Get particular about AI norms: when it’s nice to make use of, when it isn’t, who has ultimate say. Imprecise insurance policies received’t maintain up in follow.
- Some approvals exist purely as a result of issues used to maneuver slower. Verify which of them nonetheless earn their place.
- Push AI coaching previous prompting: The actual ability hole now’s realizing when to belief what AI provides you and when to push again, not how you can write a greater immediate.
Notice: This text was created by Erica Farmer in collaboration with ChatGPT 5.6 on 22/07/2026, which supported the event and refinement of her unique concepts.
Fireplace Hydrant and the O ring metaphor: credit score to Dexter Winters and used with permission.
