Eva Russell · The Future Beckons
This is not a tooling evolution. It is an operating model shift.
Most organisations are asking how to use AI. The more important question is what they need humans for now.
These are not the same question — and confusing them is the most expensive mistake currently being made.
You cannot automate your way to a better operating model. You have to build the model first.
The argument
What I think.
These are positions. Some people disagree with them. That is the point.
- 01
Layering AI onto legacy workflows doesn’t transform them. It accelerates their dysfunction.
The organisations creating real advantage are redesigning how work happens — not deploying tools on top of broken processes. You cannot automate your way to a better operating model.
- 02
Structure before intelligence. Intelligence before automation.
AI effectiveness is directly proportional to the structure of the artifacts it operates on. If your workflows are inconsistent, AI will synthesise that inconsistency at scale — faster.
- 03
“% time saved” surveys are not a credible productivity metric. They are a measure that cannot withstand scrutiny.
Self-reported time savings measure sentiment, not productivity. You can only benchmark AI impact when comparing like-for-like work — and those scenarios are rare in year one.
- 04
The real opportunity is not efficiency. It is autonomy.
When managers spend most of their week on coordination, the problem is not that they are slow — it is that they are not doing the job. The value is not the hour saved. It is what someone does with it.
- 05
The deskilling risk is structural, and it is being seriously underestimated.
The cognitive friction of doing something manually is where expertise actually forms. When AI removes that friction entirely, it doesn’t just save time — it quietly erodes the skill pipeline.
- 06
Adoption metrics tell you about deployment. They tell you nothing about whether judgment has improved.
Most organisations are measuring whether people are using AI. Almost none are measuring whether the decisions they make are better. These are very different things.
The real opportunity is not efficiency. It is autonomy. The value is not the hour saved — it is what someone does with it.
About
There wasn’t a place for this argument. So I made one.
Most writing on AI and leadership falls into one of two traps — enormous opportunity with no actionable direction, or displacement anxiety without a path forward. Neither is useful if you are trying to work out what to actually change.
The opinions throughout are mine. Thought leadership that only says things everyone already agrees with is not thought leadership — it is content.
Eva Russell
20+ years in enterprise technology and transformation.
MA AI & Digital Transformation (in progress) · University of Southampton