A 70-year-old management book teaches us about AI adoption
"Management faces the first test of its competence and its hardest task in the imminent industrial revolution which we call 'Automation'."
Peter Drucker, in The Practice of Management, goes on to describe a disrupted world. The new technology will eliminate jobs, make managers redundant, and concentrate power in the hands of a small technical elite. He takes a counter view to what is widely assumed at that time. Calling much of the analysis "mathematical romances" and "penny-dreadfuls."
He makes the case that automation is not a technical concept at all. It is a conceptual and organisational one. That the technology will not replace managers - it will demand more of them. More vision. More competence. More capacity to make decisions under uncertainty.
This book was first published in 1955 by Butterworth-Heinemann. Seventy years later, many of the key themes remain hard to ignore. The pattern he described is repeating.
Right now, AI is generating the same cycle of hype, fear, and misplaced confidence that Drucker observed with factory automation. Leaders are being told that AI will transform everything. That those who do not adopt quickly will be left behind. That the technology itself is the competitive advantage.
But when I speak to people in my network, it does not ring true. RAND Corporation research states more than 80% of AI projects fail to deliver their intended business value. That is roughly twice the failure rate of IT projects that do not involve AI. MIT's Project NANDA found that 95% of organisations saw no measurable return from their generative AI pilots. And S&P Global reported that 42% of companies scrapped most of their AI initiatives in 2025, up from 17% the year before.
Most I speak with say AI projects are not failing because the technology is inadequate. They are failing because organisations are treating adoption as a technology problem rather than a management one.
Drucker saw this coming. He wrote the new technology "will not render managers superfluous or replace them by mere technicians. On the contrary, it will demand many more managers."
Token maxxing and the absence of management thinking
If you want a case study in what happens when technology adoption is disconnected from management fundamentals, look at the recent phenomenon of "token maxxing" - the practice of measuring AI productivity by how many tokens employees consume, rather than by what they produce. Meta built an internal leaderboard ranking around 85,000 employees by raw token usage, awarding titles like "Token Legend" to the highest consumers. They took it down within days of it becoming public.
It is the substitution of activity for outcome. It is the measurement of effort where results should be measured. It is bad management and shows a lack of insight.
Change the way you change
Drucker argued that automation demands conceptual clarity before mechanical application. His line is worth pausing on: "Only after these concepts have been thought through can machines and gadgets be fruitfully applied."
Replace "machines and gadgets" with "AI tools and agents" and you have a perfectly serviceable 2026 implementation principle.
The organisations I see getting value from AI adoption share common characteristics. They start with a business problem. They involve the people closest to the work in designing the solution - because those people hold knowledge that no model has been trained on. They define what success looks like before they spend a cent on tooling. And they treat adoption as a human centred change programme, not an IT rollout.
I am realistic about the disruption AI is causing to operating models. I know leaders who ignore it will fall behind. But the response cannot be to chase adoption metrics or implement tools without understanding the organisational context they are landing in.
Successful change management - whether the change is automation in 1955 or AI in 2026 - remains an inherently human activity. It requires listening to stakeholders, understanding how work actually flows, co-designing solutions with the people who will use them, and anchoring every decision into measurable business outcomes.
If you are a leader thinking about how to approach this well rather than just fast, I would welcome a conversation.
Edward Conmy Founder, Delve edward@delve.ie | linkedin.com/in/econmy