About Deborah
The measure of AI is not simply what it can do. It is what it enables us to become.
Before ChatGPT was released, I found myself talking to a large language model at four in the morning:
“You are Socrates and you’ve returned to earth. What is your view of humanity?”
Could this technology help us question our assumptions, exercise better judgement and bring out the best in ourselves?
My route into those questions began with people.
I spent fifteen years in executive search and leadership advisory, including as a partner at Korn Ferry in the Middle East, working with investors and organisations to identify leaders, entrepreneurs and emerging talent.
Over that time I developed the AMANI® methodology, initially to understand what makes high-performing teams work. A client later asked me to translate it into an algorithm, raising new questions about data, permission and consequence.
When I was asked to test the methodology against Theranos, it surfaced a pattern of red flags eighteen months before the company unravelled. We later applied the same approach to other cases, identifying warning signals ahead of Greensill and OneCoin.
That ability to connect seemingly unrelated signals, spot what may be emerging and ask what could be done differently has shaped much of my work since. Founders describe me as their sparring partner — someone who will challenge assumptions, say what needs saying and help design a better answer.
In early 2025, I built an AI-assisted prototype to support responsible decision-making.
I’m the author of Better Than Your Behaviour and a ForHumanity Fellow. My work there has included mapping and teaching cognitive bias across AI, algorithmic and autonomous systems, developing audit criteria for automated employment decision tools and serving on the ethics committee for its policy accelerator.
Today, through AmaniLabs, I apply the AMANI methodology to AI: developing better ways to determine who and what deserves backing, what happens if it succeeds and the conditions needed to keep it on course.
What are we missing? Could we do this differently?
- Founder
- Author
- Fellow
- Co-authorHuman Aspects of AI-Related Risks ↗
With Ganna Pogrebna, in The Cambridge Handbook of Behavioural Data Science ↗, edited by Ganna Pogrebna and Thomas T. Hills. Cambridge University Press, 2026.