Document search engines will be able to think and reason like people, argues AI expert
By turning sentences into vectors, computers will be able to learn meaning for themselves, Professor Geoffrey Hinton tells The Royal Society
Machine learning is going to be very important for the future of document processing to such an extent that eventually computers and search algorithms like Google's could be able to "think like people" when examining text for meanings, ideas and thoughts.
That's according to Geoffrey Hinton, professor of computer science at the University of Toronto and a machine learning and artificial intelligence expert for Google. Professor Hinton made the remarks during his keynote speech at a machine learning conference hosted by The Royal Society today.
Professor Hinton focused much of his presentation on artificial neural networks, a model of artificial intelligence based on biological neural networks like the human brain - although he said machine learning hasn't got to that point where it can mimic human thought, yet.
However, there are already machines that learn from statistical models, with Hinton citing Netflix, which uses big data and algorithms in almost every decision, as an everyday example of this.
Voice recognition software built into many mobile phones has also improved markedly in recent years, with machine learning algorithms now reducing mistakes and errors to fewer than five per cent, which is about the same margin of error a human would make when listening to audio dialogue - and it is still improving.
Therefore, Hinton explained, if these models can be transferred to other tasks - such as more easily rendering meaning from documents for the purposes of search engine algorithms - then it could get to the point where in order to fulfil its designated task, a machine would start thinking like a human.
"The implications of this for document processing are very important. If we convert a sentence into a vector that captures the meaning of the sentence, then Google can do much better searches; they can search based on what's being said in a document," Professor Hinton told the audience at The Royal Society.
"Also, if you can convert each sentence in a document into a vector, then you can take that sequence of vectors and [try to model] natural reasoning. And that was something that old fashioned AI could never do," he continued, explaining that this means machines could potentially teach themselves to think like people.
"If we can read every English document on the web, and turn each sentence into a thought vector, you've got plenty of data for training a system that can reason like people do," Professor Hinton said.
"Now, you might not want it to reason like people do, but at least we can see what they would think.
"What I think is going to happen over the next few years is this ability to turn sentences into thought vectors is going to rapidly change the level at which we can understand documents," argued Professor Hinton.
"To understand it at a human level, we're probably going to need human level resources and we have trillions of connections [in our brains], but the biggest networks we have built so far only have billions of connections. So we're a few orders of magnitude off, but I'm sure the hardware people will fix that," he said.
Hinton believes that progress in machine learning will be driven by highly focused and curious individuals, rather than by government funded research with an end goal in mind.
"For deep learning, it's very clear that government funding encouraging translational research, if anything, slowed it down. What was important was curiosity driven research funded by, in this case David Sainsbury, not when he was Minister of Science, but in his private foundation," he said.
"In this case, the big long term advances come from breakthroughs you just wouldn't have thought of... The people who are best at coming up with those are curiosity driven scientists," Hinton concluded.
While Professor Hinton and other machine learning experts are excited by what machine learning could mean for increased productivity, there are those who fear that the introduction of artificial intelligence into the workplace could lead to mass job losses.