The Darn Messy Middle of AI
Same week. Slightly more than twenty conversations this time round. Both messy and clarifying at the same time.
Week 2 of my life, post-MAS. (Week 1 here). Still catching up on old and new connections. But I have to admit, starting to itch, just slightly, for action.
The familiar. An old friend of decades who is now CIO at a property group. The head of data at a large bank. The head of enterprise AI at another bank. A senior figure at a national AI organization who has been a great sounding board. A colleague from MAS days, who has journeyed through multiple startups. Another colleague from MAS who helped me during my PhD years ago. An ex-colleague from my art policy days now doing her doctorate in design. A recruiter at a global asset manager I knew from school who wanted to catch up over art. A director who’s spent years building adult education pathways who I met while on my last adventure. A founder building AI for social listening. A CRO at an investment firm whose path has crossed mine at every career turn.
New connections. The president of a major association in the financial sector. A professor working on cutting edge fraud detection models . The director of a university’s academy. A Gen AI lead at a global insurer. A government AI policy lead. A PhD candidate researching AI and cybersecurity who wanted to understand my weird PhD journey. A startup founder building AI for treasury. Another aspiring founder thinking about government procurement.
Totally different rooms and life experiences. But one recurring theme.
The Darn Messy Middle
2-3 years after ChatGPT, awareness isn’t the problem anymore. In fact, it’s the problem. Setting unrealistic expectations which can’t be squared.
People are increasingly aware of that strange gap between knowing and doing. A messy middle where awareness hasn’t converted to real understanding and action.
The senior figure at the national AI organization put it simply: “There’s still a huge gap between use and real understanding.”
A Gen AI lead at an insurer shared that they’re trying to deploy agentic AI for a key workflow. But the governance and risk management hurdle is still huge. It made me reflect on my own work writing the AI risk management guidelines for Singapore’s financial sector. Did I make expectations clearer, or was I part of the hurdle?
The heads of AI and data at banks who I knew well spoke of issues that we already discussed more than a year ago. Things have certainly moved. Substantially in some areas. But the progress is not easy, and hard earned. Certainly not from lack of trying.
The Takeaway Test
A director who builds adult education programmes reframed how I think about this.His philosophy: an AI course should let the learner take something back to work. Immediately. Not theory. Not awareness. Something usable. The very next day.
So simple sounding, but darn hard. Most AI content is still designed for the awareness problem. Explainers. Overviews. Frameworks. But the demand has moved downstream. People want the first step, not the big picture.
The founder building AI for social listening told of significant demand for AI training that was real. Something far beyond the current norm.
Alternative paths with AI exist. Banks using AI for simulation as training. Asset managers that are reaping huge benefits from third-party AI helping make sense of investment and risk analytics.
The Quiet Builders
Meanwhile, some are just grinding and building..
The CRO told me his firm got two interns to build an internal intelligence platform in weeks. Cloud tools. Off-the-shelf components. Done. Because AI has accelerated the process.
The professor is building foundational models for fraud detection. Not LLMs. Not chatbots. Narrow, specialized systems for specific problems. Quietly.
A PhD candidate researching explainability for fraud detection. Another layer of the same quiet work. He’s also a mid-career PhD. And reminds me of 6 years ago, when I was in the same position. Struggling, but also seeing a whole new world.
Life Goes On
And then there are rooms where AI hasn’t landed at all.
An old friend manages investments for a property group. We talked for an hour. AI barely came up. Different industry. Different pace.
A recruiter at a global asset manager from the same junior college wanted to catch up over art. We talked for a while. AI barely surfaced there either.
Same city. Same week. Different conversations entirely.
Week 2. The pattern sharpens. I’m starting to want to get back into the action.
The need for AI awareness is old news. What comes after is less clear than I expected.
What’s helping you move from knowing to doing?
#AI #AIRiskManagement #Transitions #Explorations