Most AI transformations stall for reasons that have nothing to do with the technology. In this Gamechanger Interview, Prof. Joon Nak Choi of the Hong Kong University of Science and Technology joins Victor Filamor, Partner at Stanton Chase Greater China, to explain why people, processes, and power dynamics decide whether AI investment pays back. He sets out why automating a single step captures little of what AI can do, and why rebuilding a business around AI runs into organizational imprinting, where the way a company was set up at the start turns out to be very hard to change. He then walks through the J-curve of AI adoption, which explains why 81.8% of the 214 executives in Stanton Chase’s survey could not yet report a positive quantified return, and why good change management shortens the dip. He separates risk, made up of known unknowns that a CFO can price, from uncertainty, made up of unknown unknowns that no CFO or chief compliance officer will approve, and describes the AI Workforce Engine he built with Ajay Manissery Konchery as a way to move questions from the second category into the first. Of the several hundred organizations he has advised, about two came at AI from strategy rather than from fear of missing out. He also expects AI valuations to crash within a year, and argues that this should change nothing about how leaders plan. His closing point is that companies will not so much collaborate with AI as command it, which makes the people, the culture, and the processes around it the real work.
The technology is increasingly ready. The people, processes, and power dynamics around it are not.
In this Gamechanger Interview, Prof. Joon Nak Choi, Adjunct Associate Professor at the Hong Kong University of Science and Technology and co-creator of the AI Workforce Engine, joins Victor Filamor, Partner at Stanton Chase China and Hong Kong. He draws on several hundred corporate engagements and on Stanton Chase’s survey of 214 executives across 45 countries.
In His Own Words
- On why companies start: “What’s driving AI adoption is largely FOMO, fear of missing out. You should be looking at your needs, at the pain points you’re going to solve, and then build technology for that purpose. Instead, it’s ready, fire, AIm.”
- On what actually blocks adoption: “The technology now exists. Some of it is even good. And yet they [companies] can’t get people to use it. It’s the organizational processes, the people, and the power dynamics that are preventing it from being adopted.”
- On why organizations are so hard to change: “It’s probably easier to build something new outside of your usual processes than to try to change things on the inside. This is what we call organizational lock-in, or organizational imprinting. The way you are set up at the beginning is sticky.”
- On what the returns really look like: “Everybody thinks that as soon as you implement AI, you can achieve positive ROI. But even with fairly routine automation, you are probably looking at negative ROI, because it requires change, it requires retraining.”
- On how leaders should think about the work: “I don’t like this human-AI collaboration theme. Everybody talks about it, but that’s the wrong way of thinking about it. You will have humans commanding AI.”
Why Do AI Projects Fail When the Technology Works?
Plenty of what companies build with AI is good. It still sits unused. Choi points to organizational processes, to people, and to the power dynamics inside the company. None of those move as fast as the technology does.
Why AI Stalls
The technology is increasingly ready. The organization around it is not.
Technology
Models, tools, and infrastructure keep getting more accessible, and a new generation of graduates arrives trained to build them into organizations.
People
Talent, training, trust, and whether anyone will use the available tools
Processes
Workflows, governance, and the data underneath them
Power Dynamics
Who gains, who loses, and who gets to decide
AI transformations stall on the right side of this picture, not the left.
Why Does Productivity Fall Before AI Pays Off?
AI adoption follows a J-curve. Productivity drops first, because the company has to retrain people, redesign workflows, and run the technology through several cycles before it works. Stanton Chase’s survey found that 81.8% of executives could not yet report a positive quantified return. Choi reads those missing returns as being on schedule, and says good change management shortens the dip.
The J-Curve of AI Adoption
Productivity falls before it rises, and change management shortens the dip.
Use the switch below to remove change management and see the difference.
81.8% of the 214 executives Stanton Chase surveyed could not yet report a positive quantified return on AI
What Is the Difference Between AI Risk and AI Uncertainty?
Economists use the word risk narrowly. Risk means known unknowns, the things you know you do not know, and you can put a number on them. Most of what gets called AI risk is uncertainty, which is made up of unknown unknowns. Nobody can model what has not been identified, so CFOs and compliance officers will not sign it off.
Turning Uncertainty Into Risk
Send a question for investigation and no answer comes back. What comes back is a list of things your team can measure, which is what makes the spending approvable.
Click any red card on the right to send it for investigation.
Known unknowns
You still do not have the answer. You know what the answer would consist of, which is enough to scope and price the work.
The CFO can approve it.
Unknown unknowns
Real worries, but nobody has put a number on them, so there is nothing to price.
The CFO and compliance cannot approve it.
Five open worries, none of them measured. A CFO has nothing to price, so nothing gets approved.That did not settle the question. It worked out what settling it would take, and what that costs.The same five worries, still unanswered. Each one now has a set of measurements your team can take, and that is enough for a CFO to price the work and sign it off.
How Do Leaders Get AI Investment Approved?
The work is turning uncertainty into risk. Choi says it starts with listing the questions the organization needs to answer, then sorting them into what is known and what is not. That is what the AI Workforce Engine™ is built to do. He created it with Ajay Manissery Konchery.
How Many Companies Start with an AI Strategy?
Of the several hundred organizations Choi has worked with, about two came at AI from strategy first. The rest are reactive, driven by fear of missing out and by boards watching what competitors are doing. He calls the result “ready, fire, AIm.”
What Should CXOs Do About AI Now?
Choi expects an AI crash within a year, then much larger returns for the companies that keep working through it. His advice does not change either way. Decide what the organization is trying to achieve, show the process before and after, and plan for the people who will command the AI.
Watch the full interview here.
Watch More
Read More
How Can We Help?
At Stanton Chase, we're more than just an executive search and leadership consulting firm. We're your partner in leadership.
Our approach is different. We believe in customized and personal executive search, executive assessment, board services, succession planning, and leadership onboarding support.
We believe in your potential to achieve greatness and we'll do everything we can to help you get there.
View All Services