Ep. 218: Nuno Reis, PhD - On Uncertainty, The Dangers of Probability Dogma, and more
May 17, 2024
auto_awesome
Nuno Reis, a PhD expert in uncertainty and probability, criticizes Bayesian Thinking's dogma in AI models. He relates his experiences in the financial sector during the Global Financial Crisis to current parallels in AI. The podcast covers Bayesian Thinking, uncertainty, philosophy, and running, offering insightful perspectives on embracing uncertainty and creativity.
Bayesian Thinking can become dogmatic, hindering alternative modes of thinking.
Over-reliance on quantified models restricts creativity and diverse thinking patterns.
Encouraging diverse thinking through methods like improvisation and lateral thinking is crucial for personal growth.
Deep dives
Bayesian Thinking Overview
Bayesian thinking originated in the 17th century through the work of statisticians like Thomas Bayes, focusing on inverse probability and conditional probabilities. It allows for beliefs and subjective experiences to influence decision-making based on personal understandings. The technique aims to update beliefs through experience, enhancing decision-making processes over time.
Challenges of Bayesian Thinking
Bayesian thinking can become dogmatic when subjective beliefs are taken as absolute truths, leading to self-fulfilling prophecies. There can be a shift from objective, mathematically based knowledge to subjective belief-based interpretations. This shift may restrict creativity and alternative modes of thinking, constraining individuals within a quantified rationality framework.
Limitations of Quantified Models
Quantified models, like Monte Carlo analysis, come with limitations that may not be solved by increasing model accuracy. Understanding the constraints of models and the need for diverse thinking is crucial. Over-reliance on quantified approaches can hinder lateral and creative thinking, constraining individuals within specific thinking patterns.
Exploring Diverse Thinking Approaches
Encouraging diverse thinking involves exploring techniques beyond quantified models, such as improvisation, dream analysis, or lateral thinking. These methods offer alternative perspectives and creative avenues for decision-making. Embracing a broader range of thinking styles helps individuals break free from constrained and limited thinking patterns.
Exploring Creativity and Lateral Thinking
The podcast delves into the importance of creativity and lateral thinking, emphasizing the value of curiosity and questioning in one's approach to work and personal growth. It highlights that exploring uncertainty and embracing curiosity are vital for enhancing creativity and generating new ideas. The discussion underscores the significance of allowing moments of boredom and silence to foster creativity, contrasting with the constant stimuli of social media and overstimulation.
University of Uncertainty and Resilience
The conversation shifts towards the concept of the 'University of Uncertainty,' designed to help individuals navigate uncertainty and connect their insights for better sense-making. It is presented as a program rather than a structured curriculum, focusing on personal exploration and understanding rather than traditional goal-oriented education. Encouraging participants to question dogmas and understand their internal expertise, the initiative aims to challenge conventional thinking and foster adaptive responses to complexity.
Today, I have the pleasure of welcoming Nuno Reis to the show. I came across Nuno via his LinkedIn posts on uncertainty and in particularly around something called Bayesian Analysis or Bayesian Thinking. If you’ve never come across the term, Bayesian Analysis is the mathematical interpretation of probability. And it underpins so much of our world - and increasingly so because many AI models are built on Bayesian Thinking.
Nuno is quite critical of how Bayesian Thinking is applied – because we can never remove the human from the numbers and the models. He says that an embrace of Bayesian Thinking – useful as it is, has become an unhealthy dogma. And – that resonated with me particularly because as I looked at Nuno’s background I saw someone saying this who is a trained mathematician with a PhD in Mathematics in String Theory and did a postdoc at Oxford. He worked in the financial sector during the Global Financial Crisis where he sees parallels now in the worlds of AI and related areas.
So, I invited him to the show and we had a great conversation that covers not only the topics of Bayesian Thinking, the financial industry, but also the topic of uncertainty, lots of philosophy and running. I hope you enjoy it and find it insightful. Here is Nuno Reis.
