That’s right – the Nobel of Computation goes to trio known as ‘the godparents of AI’.
You may not know Yoshua Bengio, Geoffrey Hinton, and Yann LeCun, but you may have already interacted with one of your “puppies” today.
The trio is a pioneer in artificial intelligence (AI), and it is exactly because of work like theirs that we can now have facial identification systems or so accurate translation and pronunciation in several languages.
For this reason, the “godfathers of the AI” received the Turing 2018 Award, the “Nobel Prize in Computer Science,” and they share the US $ 1 million – and, of course, they are awarded as key development agents of machine learning.
The techniques developed by Bengio (55 years old), Hinton (71) and LeCun (58) in the 1990s and 2000s allowed for significant advances in tasks such as computer vision and speech recognition.
His work supports, for example, technologies ranging from stand-alone cars to automated medical diagnostics.
The “winter of AI.”
Although neural networks and machine learning are commonplace in today’s technology, they were simply known as AI and lived in cycles of popularity oscillation.
When expectations were not met, investment in the industry was frozen – and this was known as the “AI winter.”
This created a small community that, by 2012, 2013 actually.
It was at the end of one of those periods in the late 1980s that Bengio, Hinton, and LeCun began to exchange ideas and work on similar issues.
One was just about computer programs made from connected “digital neurons” – the neural networks, which became the fundamental foundation for modern AI.
The trio decided they needed to rekindle the interest of all got funding from the Canadian government to sponsor an interrelated research center.
“We organize regular meetings, regular workshops and summer schools for our students. This created a small community that, by 2012, 2013 did mesh, “recalls LeCun.
They showed excellent results in tasks such as character recognition, and in 2012 they caught the attention they wanted when a group led by Hinton took over ImageNet.
The project is merely the fundamental pillar for deep learning in image recognition.
And where are we going?
With hardware advancements, such as increased processing power and lower cost on graphics cards and plenty of digital data with the improvement of the internet, all the cognitive mechanisms introduced by the trio have spread and evolved – they have become ubiquitous in most devices we use daily.
We do not have machines with common sense, LeCun says.
LeCun says he is optimistic about AI’s prospects but makes it clear that there is a lot of work to be done.
Current systems need a lot of data to understand the world and can be easily fooled and good only at specific tasks.
“We just do not have machines with good sense.”
If the field continues in the direction of today, new methods will need to be discovered, as fundamental as those developed by “AI sponsors.”
“Even if we can use new techniques to create intelligence on a human level, there will be another 50 mountains to climb, including those that we still can not see.”