93% of AI Investment Is Going Into Technology. The Harder Part Is the People Around It.
By Malin Gaertig · September 29, 2026
93% of AI investment is going into technology. Just 7% is going into people, process, culture and reskilling — and the strongest predictor of AI adoption isn't a growth mindset, it's whether people are exposed to leaders who actually use AI themselves.
- —93% of AI investment is going into technology. Just 7% is going into people, process, culture and reskilling.
- —Organisations are spending on tools and agents, but the redesign of the work around them is lagging behind — Deloitte's Ben Fish on the gap he sees with clients.
- —The strongest predictor of AI adoption wasn't a growth mindset — it was whether people were exposed to leaders who actually used AI themselves.
- —AI adoption is not just a training problem: people watch what leaders do, and visible use gives them something to copy.
- —As AI makes decisions about which products, brands and information customers see, someone must answer: what decision is it making, what evidence is it using, and who is responsible for understanding the result?
| Spending on tools | Changing how people work | |
|---|---|---|
| Share of investment | Technology — 93% of AI investment. | People, process, culture and reskilling — just 7%. |
| What leaders assumed predicts adoption | A growth mindset across the workforce. | Whether people are exposed to leaders who actually use AI themselves. |
| How the message travels | The leadership team talks about AI but doesn't use it — the organisation gets a message. | Leaders use it openly, show how, and make that behaviour visible — something people can copy. |
| The question to answer | Which tools should we buy? | What decision is it making, what evidence is it using, and who is responsible for understanding the result? |
93% of AI investment is going into technology. Just 7% is going into people, process, culture and reskilling.
That was one of the numbers that stayed with me from the ARC AI Summit. The technology is increasingly easy to buy, but the harder (and less funded) part is changing how people work around it.
On the leadership and workforce panel, Deloitte's Ben Fish described the gap he is seeing with clients: organisations are spending on tools and agents, but the redesign of the work around them is lagging behind.
What predicts AI adoption better than a growth mindset?
Then came a surprising insight: one bank had assumed that growth mindset would be the strongest predictor of AI adoption. However, when they tested the hypothesis against their workforce data, the strongest predictor was whether people were exposed to leaders who actually used AI themselves.
Why does leader behaviour matter more than training?
That is a very different answer because it suggests AI adoption is not just a training problem, but rather people watch what leaders do. If the leadership team talks about AI but doesn't use it, the organisation gets a message.
If leaders use it openly, show how they use it, and make that behaviour visible, they give people something to copy.
What questions should someone inside the business be able to answer?
The same principle applies beyond internal adoption. As AI starts making decisions about which products, brands and information customers see, someone inside the business needs to understand what is happening and why.
Key questions to answer are:
What decision is it making? What evidence is it using? And who is responsible for understanding the result?
How does MARK fit into closing this gap?
That's what MARK is about — it measures whether AI selects your brand and products and what evidence appears to drive that selection, so the answers to those questions aren't left to assumption.
The 93/7 split says the technology is bought before the understanding is built. Agentic Readiness Score shows how ready the organisation is for agentic commerce, and Share of Model shows what customers' AI models actually decide about you — measured, not assumed, the same distinction Bunnings applies to its own assistant in Did AI make the right decision?