‍

There are many pioneers in the field of artificial intelligence who have broken new ground, but none can match the remarkable career of Yoshua Bengio. His relentless pursuit of new machine learning techniques has led to him being known as one of the three “Godfathers of AI”, but he has been careful to couple his remarkable discoveries with a determination to ensure they are used for the good of humanity as a whole. He’s not just a pioneer, but an influential voice guiding the future development of the technologies he has brought into the world, a True Superhero of ethical as well as artificial intelligence.  

‍

‍

From Hippy Son to Tech Godfather

Yoshua was born in 1964 in Paris, France, to Moroccan parents. His father, Carlo, was a pharmacist and his mother, Celia, was an economist, but neither of them worked in these fields. Both were hippies, caught up in the rush of the monumental social change of the 1960s, and their interests hewed closer to the arts than the sciences that were their specialties; they worked in small community theaters which led, by necessity, to a frugal life. This did not bother the Bengios, who were active participants in the 1968 Revolts of Paris and the revolutionary community that sprang up in that city, but it did mean that Yoshua and his younger brother Samy had to work themselves to afford the primitive hardware that sparked their careers in computing. The boys pooled the funds they had earned from newspaper rounds in order to purchase an Apple II and an Atari 800, and these machines provided the platform for their first experiments in computer programming as teenagers as the 1980s began.  

‍

‍

Those first experiments took place in Canada, not France, as the Bengios had emigrated to Montreal in 1977, after spending a year in Casablanca while Carlo completed his military service. Both Yoshua and Samy (now also a leading figure in artificial intelligence) were fascinated by the world of computers and programming, and their studies took them to McGill University in Montreal, where Yoshua completed a bachelor’s degree in computer engineering in 1986. While studying for his degree, Yoshua was introduced to the concept of neural networks - computational systems that attempt to mimic the functionality of the brain   - and the work of Geoffrey Hinton, who would later share the Turing Award with Bengio and the French AI pioneer Yann LeCun. Bengio was particularly inspired by Hinton’s work exploring the nature of human intelligence, and began to focus his studies on this burgeoning new field.

‍

A pioneering career

Neuroscientists had been working to try to create neural networks since the 1950s, but the field had stagnated as researchers reached the limits of the hardware available to them, and very little progress had been made since the 1960s. The prevailing wisdom within mainstream computer science circles was that artificial intelligence belonged firmly in the realm of science fiction. Yoshua had been an enthusiastic reader of sci-fi ever since he was a child, and did not share the skepticism of the wider community. In 1991 he completed his Ph.D. on neural networks, and swiftly set about deploying his research in real world applications. As a grad student he met Yann LeCun for the first time, and they collaborated on a project derived from the ideas in Yoshua’s Ph.D. thesis to develop a system for handwriting analysis, which was used by AT&T to develop automated systems for processing paper checks. This quickly revolutionized the banking industry, which had previously required several days to carry out the “clearing” process of verifying checks. 

‍

‍

Yoshua remained in academia, returning to Canada in 1993 as a faculty member at the University of Montreal, where he began work on the next phase of his pioneering career, developing groundbreaking language processing techniques. In 2000 he published his famous paper, “A Neural Probabilistic Language Model”, which radically altered the way computers understand and process human language. This led to an enormous number of technological advances that are now commonplace in our daily lives, from autocomplete suggestions on your phone to automatic language translation services and beyond. 

‍

Bengio’s next breakthrough came in collaboration with one of his Ph.D students, Ian Goodfellow; together they developed the concept of “generative adversarial networks”, where  two networks compete against each other in order to achieve the best result at a given task, a method inspired by the concept of game theory. This process has countless applications but is best known as the method through which AIs are able to create new images from a given prompt, as popularized by services such as Dall-E and Midjourney. One network attempts to create a new image, and the other attempts to detect whether it is a fake or a real, pre-existing image. As the process repeats the fakes become more and more refined, ultimately leading to remarkably convincing results. 

‍

Great responsibility 

Generative adversarial networks are controversial, as they are open to abuse. The first application of this technology to reach the media’s attention were “deepfakes”, artificially generated images and video that depicted entirely fake situations. The most infamous of these were usually pornographic and involved celebrities, and provided a stark illustration of the potential for abuse of AI technology. As one of the godfathers of AI Yoshua has strived to push back against this threat by fostering altruistic applications of his work, and has developed startup incubators that work to promote businesses and applications that seek to improve the human condition, in areas ranging from healthcare to immigration. His commitment to developing these technologies in an ethically responsible manner is just as laudable as his work in pioneering them to begin with, and as his Truly Superheroic career continues the only question is what this remarkable mind will come up with next.  

‍