Decision-Making in the AI era.... We are in for a ride!

Last week, I had the pleasure to discuss AI in Finance at a Bloomberg breakfast for leaders in Madrid. Great breakfast, fun group of people!

Few quick highlights of what we covered in just 90 mins

The real AI bottleneck is data quality, not model sophistication

Human trust and personalized advisory still differentiate in a digital world

Metadata is becoming critical for AI to find and use information

There is a global talent shortage in AI-ready financial professionals

Behavioral biases on the receiving end of AI output may be the last unsolved bottleneck

2 points to highlight and build on as they pertain to decision-making

Leaders said that they were not laying off people in their organizations for the benefit of AI. They saw the technology as a boost (augmentation) of current talent, and not a cost-cutting mechanism like often portrayed to us by media or big tech layoffs. The term Jevons Paradox comes to mind as it explains that efficiency gain in economies does not necessarily decrease the use of the product (or operator), like when fuel became more efficient, we used more of it, not less. My students may sleep better after this..

Point number 2 builds on the first one and comes from the science of decision-making. Decisions made quickly by few people (or one) are not always optimal - they are only in some cases, where a decision is inconsequential and reversible (referencing the Amazon DM Matrix). When AI is seen as a way to reduce headcount, you are leaving more decision power in the hands of a few, using AI systems which are not yet powerful enough to make agentic decisions 100% trustworthy at scale. So what happens when the AI hallucinates, which it often does, and you've outsourced your big decision to its probabilistic logic? Governance frameworks do not yet allow proper responsibility to be assigned to a wrong decision made by AI - hence the term human-in-the-loop.

The question is: as monetization of agentic AI becomes more demanded, what will happen to the accountability of good/bad decisions in organizations? Who is responsible when firms outperform the market via AI? What if things go wrong?

Thank you for the opportunity to join. It was fun to learn how you're navigating the space.