The Theory of Anything

When Alpha Go beat Lee Sedol, the world Go champion, it came up with creative new moves never previously seen before and even invented a whole new style of play unknown to humans. IBM's Deep Blue, the champion chess algorithm, failed to do either of these. What was the difference? In this podcast, we review Alpha Go the Movie. Warning: Spoilers abound! Please go watch the movie first! This is an excellent movie.  Bruce (using his admittedly thin knowledge of reinforcement learning) explains how Alpha Go works (using the materials previously discussed in our Reinforcement Learning episode) and how Alpha Go came up with a creative new approach to Go that went beyond the knowledge of the programmers.  While Alpha Go definitely does not have "creativity" in the universal explainer sense of the word (it has no explanatory knowledge nor understanding), it did come up with a creative new playstyle never before seen in the history of the world that changed how humans play Go. Even the programmers were caught off guard by what it came up with. We talk about how Alpha Go challenges the Pseudo-Deutsch Theory of Knowledge but meshes well with Campbell's evolutionary epistemology. 

We wrap up our discussion with Sadia Naeem covering possible solutions and criticisms of those solutions. 

Sadia Naeem continues the discussion about unsolved problems in physics. This time we talk about (among many other things) symmetry and novelty. 

Sadia Naeem joins us again, this time to explain clocks, block universes, and eternalism. 

A physicist discusses the problems and mystery of time, including string theory, emergence, and the illusion of time. The concept of pseudo randomness and the exploration of free will and determinism are also covered. The podcast challenges the traditional status quo in physics and highlights the importance of the problem of time.

In this episode Cameo, Tracy, and Bruce geek out over how good the Marvel TV shows are and how much they really get right. Spoilers abound, so be warned. 

Learn about reinforcement learning and its stunning victory in Alpha Go. Discover the theory of reinforcement learning and how it solves Markov Decision Problems. Explore important concepts such as state transition and reward functions. Find out the origins of dynamic programming and its challenges without knowledge of these functions. Gain insight into key concepts in reinforcement learning and its potential as a general learning algorithm. Lastly, delve into the learning abilities of animals and comparison to humans and machine learning.

In this episode, we interview Chiara Marletto about her recent book The Science of Can and Can't: A Physicist's Journey Through the Land of Counterfactuals as well as discussing Constructor Theory in general and how it might help us form a new mode of explanation in physics. We ask her some tough questions about constructor theory and she fields the questions very well.  For those interested in q-numbers vs real numbers, see Sam Kupyer's lecture on our Youtube channel. Follow us on Twitter. Check out our blog.

This episode examines whether the Universal Darwinism algorithm applies to knowledge creation in AI and machine learning. It explores examples like Gradient Descent and the Naive Bayes Classifier, questioning their relationship with variation and selection. Counter examples challenge Donald Campbell's prediction about evolutionary algorithms being the sole source of knowledge creation. The podcast also delves into the theory of critical rationalism and the value of testable explanations. Overall, it highlights the exciting epistemological problems that Machine Learning presents.

The podcast explores the debate between David Deutsch, Karl Popper, and Donald Campbell regarding knowledge creation in evolutionary algorithms. It discusses the overlap between artificial intelligence and Popper's epistemology, the steps of the algorithm for improving solutions, the omission of the third step in blind variation selective retention, and the creation of knowledge through variation in selection. It also delves into competing views on knowledge creation, blindness in variation and selective retention, and the concept of blindness in the Universal Darwinism algorithm.