Conscious or reactive: which kind of AI adoption are you running?

You sit in the quarterly review and someone asks what the AI programme is actually for. You have the copilot licences, the pilot in customer service, the agent everyone is excited about, the second pilot nobody quite owns. And you hear yourself say it: we keep bolting on whatever tool is in the headlines, and I cannot actually tell you what we are trying to become.

That sentence is the whole diagnosis. There are two kinds of AI adoption, and the difference between them is not budget, model choice, or vendor. The difference is direction. Reactive adoption starts from the headline: a competitor announces something, a board member forwards an article, and a tool gets bolted on in the hope that a strategy will assemble itself later. Conscious AI adoption starts from a deliberate intent, a clear statement of what the business is becoming, and then selects the technologies that serve that intent. The reactive path produces activity. The conscious path produces transformation. They look identical on a spend report and could not differ more in the numbers a year later.

Why does my AI programme feel busy but not transformational?

Because busy is what reactive adoption manufactures. The behaviour is now the norm, not the exception. In IBM's 2025 study of 2,000 chief executives across 33 countries, 64% acknowledged that the risk of falling behind drives them to invest in some technologies before they understand the value those technologies bring. IBM's 2024 study put that figure at 51%, so the pattern is real and intensifying. The phrase the coverage attached to it was blunt: much of the spending is driven by the fear of missing out, not by return.

This is the gap that makes the quarterly review feel hollow. You are moving quickly, and quick feels like progress. But movement toward an undefined destination is just motion. The headline supplies the urgency; it cannot supply the intent. So the tools accumulate, each one defensible in isolation, and the portfolio as a whole answers a question nobody asked. You can feel the dissonance before you can name it, which is exactly what that sentence in the review was.

What does the reactive pattern actually cost?

It costs the return. In that same IBM study, only 25% of AI initiatives had delivered the expected return over the previous few years, and just 16% had scaled across the enterprise. Deloitte's 2025 report, AI ROI: The Paradox of Rising Investment and Elusive Returns, surveyed 1,854 senior executives across 14 countries and found the same shape from the spending side: only about 20% qualify as AI ROI leaders, just 15% of generative AI users report significant, measurable return, yet 91% plan to increase their AI investment over the coming year. Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027, citing escalating costs, unclear business value, and a market thick with what it calls agent washing (vendors relabelling ordinary automation or chatbots as autonomous "agents").

Read those numbers together and a clear shape appears. The spending is real, the activity is real, and the transformation is largely absent. That is not a technology failure. The models work. The failure is one of intent, dressed up as a tooling decision.

What separates the few who get a real return?

Not the algorithm. PwC's 2025 Global AI Jobs Barometer, built on close to a billion job adverts across six continents, found that productivity growth has nearly quadrupled in the industries most exposed to AI, rising from 7% in the years to 2022 to 27% by 2024, while the least exposed industries flatlined near 9%. Revenue per worker is now growing roughly three times faster in the most AI-exposed industries than in the least. The advantage tracks the skills and the redesigned work, not the model itself. Wharton's 2025 study, Accountable Acceleration, surveyed more than 800 senior decision-makers at US firms of a thousand staff or more and found the winners share a discipline the others lack: 74% now report a positive return from generative AI, and 72% formally track that return rather than assume it.

This is the part the headline can never give you. The bottleneck is no longer the technology. The bottleneck is the clarity of the human intent the technology is asked to serve, and the willingness to redesign the work around that intent rather than sprinkle a tool over the top. Conscious adoption is simply the practice of leading from where you intend to be, and letting each tool earn its place against that intent.

There is a quieter layer underneath this, for the leader who wants it. The same regulation that lets a person hold a clear intent under pressure has a measurable physiological signature. The neurovisceral integration model (a body of autonomic research linking the heart's beat-to-beat variability to how well the prefrontal cortex governs attention and decision-making) connects higher heart rate variability to stronger executive function. Read that as exploratory, and read it as a clue. Coherence in the leader, the state where head and heart align, tends to precede coherence in the strategy. The calm to define the intent before chasing the tool is itself a capability you can build.

Conscious adoption is a posture, and you can step into it this quarter. Three moves.

1. Name the intent. Write one sentence describing what the business is becoming over the next three years, with no tool named in it. If that sentence is hard to write, that is the work to do before any procurement.
2. Redesign one workflow to match it. Choose a single process that the intent makes important, and rebuild it for the outcome. The tool comes last, selected to serve the redesigned workflow rather than the reverse.
3. Measure value, retire activity. Hold every pilot against the intent and the profit it moves. The ones that serve it scale. The ones that were chasing a headline get released, and the energy returns to the work that is becoming something.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.