What does it take to be the godfather of AI? And, how does someone come to obtain such a legendary title?
In this episode of Short and Sweet AI, I talk about Geoffrey Hinton, a neuroscientist, computer scientist, and the man Google hired to make AI a reality. In many ways, we have Geoffrey Hinton to thank for developing modern AI and deep learning. It is thanks to him that deep learning has become mainstream in the field of artificial intelligence.
So, how did Geoffrey Hinton rise to become the godfather of AI?
Listen to this episode of Short and Sweet AI below or read on for more…
How Geoffrey Hinton became the godfather of AI
Geoffrey Hinton’s name is synonymous with neural networks and deep learning, but no one saw the value of his research for decades. To become the godfather of AI, he was willing to believe in himself through years in obscurity. Despite others criticizing his work, Hinton continued to work around the clock. Overcoming his peers’ harsh critics, he had faith in backpropagation, which is the workhorse of deep learning despite the AI winters of the 1990s when research funding dried up.
Hinton relocated to Canada and convinced the Canadian government to financially support research in an AI wilderness. Finally, after decades, Hinton had the technology he needed to power neural networks. This enabled him to showcase deep learning at a global competition, where he blew everyone away. That is the story of how Geoffrey Hinton became the godfather of AI.
Should machines think the way humans do?
Geoffrey Hinton wanted to build machines that think the way humans do. This became the subject of a somewhat heated debate between great minds in the AI industry. After all, is it even possible for machines to think like humans?
Nonetheless, Geoffrey Hinton was determined to build machines that can think the way humans do. He was obsessed with how the mind worked and wanted to mimic the brain using computer algorithms to create a pure form of machine intelligence. However, symbolic reasoning was the prevailing theory of how to teach machines to think. Symbolic AI uses facts and rules to train machines to manipulate symbols, and yet, Hinton disagreed.
Hinton’s discovery: deep neural networks
Research showed human learning in the brain takes place in numerous layers, with information being processed and passed on to successive layers known as neural networks.
Hinton believed neural networks were a simplified model of how the brain works. He created computer algorithms that coded for networks with thousands of layers. These layers became known as deep neural networks and gave rise to deep learning.
Deep neural networks replicate how the brain processes information. However, the AI community was not convinced and turned its back on Hinton. Many people warned Hinton to give up and put his focus elsewhere. His PhD supervisor regularly told him he was wasting his time and that neural networks wouldn’t work. His research papers were given bad reviews and were bombarded with critical comments such as “Hinton’s been working on this idea for seven years and nobody’s interested, it’s time to move on.” Hinton’s ideas were ridiculed, ignored, and called completely crazy. But he persevered.
How deep learning became mainstream
Hinton continued to receive backlash from the AI community until things began to change in the mid-2000s. GPUs, which perform millions of math calculations in milliseconds, gave computers more speed and power. At the same time, big data provided the enormous datasets required by neural networks. All of this technology combined helped to accelerate Hinton’s research. He and his team forged ahead and built powerful deep learning algorithms, including the creation of ImageNet in 2012.
Hinton’s deep learning neural networks were so powerful that they outperformed all other algorithms. The evidence was indisputable, and the validity of his research was finally accepted. It wasn’t long before students began to turn away from traditional, machine learning research to work on deep learning. Internet giants such as Google, Apple, and Facebook, embraced Hinton’s work and developed their own deep learning applications.
Today, Geoffrey Hinton is considered an artificial intelligence icon, and deep learning has become mainstream after 30 years in the wilderness.
How deep learning became AI’s “lunatic core”
Even after all of his success, Hinton continues to be a maverick and is now working on a new type of AI called capsule networks. Deep learning requires huge quantities of data to learn. To get around this problem, Hinton thinks capsule networks can lead to something called self-supervised learning which doesn’t require large datasets.
Colleagues now recognize how Geoffrey Hinton pushed forward the frontiers of neural networks and deep learning for 30 years. Some might argue that the deep learning revolution would have come regardless. However, speech recognition, computer vision, autonomous vehicles, all came sooner because of Hinton.
The table has turned, and deep learning has become mainstream. As Hinton describes it, “We ceased to be the lunatic fringe. We’re now the lunatic core.”