My undergraduate degree is in Psychology. When I first started I was fascinated by the depths of the human psyche, I read a lot of Carl Jung’s work and this book called The Denial of Death by Ernest Becker (which is really interesting, also Man’s Search for Meaning by Viktor Frankl and Homo Sapiens + Homo Deus by Yuval Harari) but after a while I felt that I needed something more concrete to supplement the abstractness of depth psychology, which is what sparked my curiosity for neuroscience and artificial intelligence (AI).

Systems neuroscience is the study of neural networks in the brain and how they produce algorithmic circuits that lead to higher cognitive functions such as vision and language processing, computational neuroscience builds mathematical models, algorithms and simulations that aim to be biologically plausible, in order to test discoveries from systems neuroscience. There are AI algorithms that are conceptually inspired by neuroscience such as reinforcement learning, there are also AI algorithms that inspire research into biologically plausible models, however, due to the complexity of the brain and intelligence, both computational neuroscience and AI require a lot more research to reach the aims of artificial general intelligence. This is one of the reasons I am dedicated to the endeavour, and why I chose to work in the intersection of neuroscience and AI; there is a lot of debate around whether neuroscience has been and is relevant for AI, and whether AI will improve beyond the latest advances, in all this debate it seems to me that the field is still in early stages, which is what makes it more exciting, as so much has been accomplished before we've reached an advanced stage. I believe the only thing missing is a focus on the limitations of the healthy human brain in terms of the networks that cause these limitations of higher cognitive functions, and how these networks could be improved; but this is part of the evolution of the field.

In the book Homo Deus, Yuval Hariri posits that consciousness could be what holds humans back from reaching higher or alternate levels of intelligence. This is what inspired me to think about the failings of the human mind and brain, as well as its successes, and how this would apply to the development of AI. This approach will allow researchers to form concepts and build AI that work better than the brain. To clarify, it is clear to me that the intersection of computational neuroscience and artificial intelligence, that is the back-and-forth hypothesization, is key to the future of this technology, in terms of understanding the structures/algorithms that produce high levels of intelligence, the mathematics and computer science are also key to forming the intricate details of these algorithms. After the understanding of higher cognitive functions in humans is solidified, which could take hundreds of years, the development of artificial neural information processing systems will take a new direction, away from human intelligence and more focused on higher levels of intelligence in artificially intelligent machines, as the literature suggests.

DeepMind demonstrates this endeavour; I will use their software AlphaGo as an example. To my understanding, AlphaGo uses many different forms of AI, one being reinforcement learning, which is inspired by the workings of the brain of many animals, including humans (https://www.nature.com/articles/s41593-018-0147-8). AlphaGo can form novel, arguably creative, ways of playing the game Go due to abilities that would be considered superhuman. I will study this algorithm further and write a post focused on its technical aspects (which will include code), but from a philosophical perspective it seems that AlphaGo was able to rapidly perform probabilistic computations to develop moves that were unlikely to be played by a human, which led to it winning most matches against one of the best Go players in the world, for the first time in history. It was able to make these decisions by studying thousands of examples. Alpha Go was also able to think ahead in order to calculate the chances of a certain move leading to its success. It must have also been able to infer information from its opponents strategy and adjust its own strategy accordingly. The most interesting thing about this algorithm is that it was not programmed to make certain moves, it was only given data and trained, therefore its perceived creativity emerged spontaneously. Though, it probably was not thinking about playing a game or that it was representing artificial intelligence, it was just computing probabilities.

The brain is often compared to a computer. I am starting to see this likeness, though, from a psychological perspective, there are features of the human psyche which separate the brain from the workings of a computer. Interpersonal neurobiology, for instance. The study of human relationships and brain development. This field explores the effects of interpersonal relationships, such as the mother-child relationship, on the physical structure of the brain; it is based on the theory that brain development is a relational process. Findings from interpersonal neurobiology are quite profound as they suggest that the brain is physically affected by non-physical interactions, which posits a complex relationship between the mind, body/brain and relationships. Though it would be interesting to study phenomena like this from a computational viewpoint (studies have shown that today’s artificial intelligence struggles to update its database through interpersonal communication/question asking https://cims.nyu.edu/~brenden/papers/RotheEtAl2019CogSci.pdf), I would not want to make a modern machine more human-like because they are entirely their own thing, in the same way that a human and a whale are different; though this is a good place to start I believe the full potential of these machines and software is still beyond our imagination and they will be very different from humans (higher levels of consciousness maybe?), which is what makes them so fascinating and what motivates me to study and work with artificial intelligence.