Geoffrey Hinton’s Nobel win product of persevering

The research that won Geoffrey Hinton a Nobel Prize for physics was the product of plenty of work carried out before artificial intelligence was the buzzword it is today.

The British-Canadian computer scientist and other AI pioneers say his now-celebrated discoveries dating back to the 1980s attracted doubters and a fraction of the attention AI sees today.

While Hinton remembers many of his research efforts as fun, he says it was “slightly annoying” that several people were skeptical of some of his theories.

He says these skeptics thought neural networks, which are models that mimic the human brain by recognizing patterns and making decisions based on data, were a waste of time and would never be able to learn complicated things.

Yoshua Bengio, a fellow Canadian computer scientist who won the A.M. Turing Award with Hinton in 2018, says it took about two decades for the perception of neural networks to shift.

He says it took so long because schools of thought can be really entrenched and difficult to change, even in the scientific community, leaving Hinton frustrated for many years as his ideas were rejected by the mainstream.

Hinton was awarded the Nobel Prize on Tuesday for uncovering a method that independently discovers properties in data and is seen as foundational for the large neural networks AI relies on. Co-laureate John Hopfield was honoured for advancing AI by creating a key structure that can store and reconstruct information.