From Musk to Putin, people are talking about the future role – and unintended consequences – of AI, but what do we really know about it?
Yesterday, Elon Musk was all over the news (again) for his latest grandiose statement, albeit one that was of a decidedly less rosy sort. “Competition for AI superiority at national level most likely cause of WW3 imo,” he tweeted.
Musk’s dire World War III warning was prompted by comments from Russian President Vladimir Putin a couple of days earlier. “Artificial intelligence is the future not only of Russia but of all of mankind,” said Putin. “Whoever becomes the leader in this sphere will become the ruler of the world.”
Perhaps, but at this stage of the game the impacts and opportunities inherent in artificial intelligence – hailed as having the potential to be a modern industrial revolution – remain undefined for even the most prominent thinkers on the subject.
This past spring, at C2 Montréal’s very first AI Forum (co-created with Montréal artificial intelligence incubator Element AI), researchers, technologists, philosophers and entrepreneurs guided us through the possibilities of what the future might hold. They also covered the basics of what we know AI is (and isn’t).
This ISN’T a takeover
Let go of your Skynet fears. According to Element AI co-founder and deep-learning luminary Yoshua Bengio, we’re still very far away from human-level intelligence.
Sam walks into the kitchen.
Sam picks up an apple.
Sam walks into the bedroom.
Sam drops the apple.
Where is the apple?
In the bedroom.
Basically, AI is a toddler
That said, research is still moving at breakneck speed. In fact, Yoshua claims that even if we stopped all research today, it would still take us more than 10 years to reap the benefits of all the progress and tools we’ve made so far.
The next AI milestone: intuition
For now, algorithms think in fairly superficial ways, requiring structured human supervision and answers in order to learn – that’s what deep learning is trying to change. Apple co-founder Steve Wozniak said he thinks humans are still a few leagues ahead.
“AI is a vague category,” Steve said under the 360 Big Top at C2. “It’s not like a human brain. It’s a super-fast specialist that follows a particular instruction. We’re very far from artificial intuition, which would replicate the uniquely human ability to select a problem to be solved and choose the methodology to be used.”
It IS capable of most human tasks
A new method associated with deep learning is now mimicking the way we teach humans – starting with sixth grade math before leaping to computational logic. It’s called curriculum learning and has led scientists like Yoshua to believe that most human tasks and knowledge could be replicated.
“There’s no reason to believe that we will not be able to understand the kind of principles that make us intelligent, and build machines based on those principles,” he said.
It ISN’T a robot
Though it may be tempting to imagine AI taking the form of your very own R2-D2, that’s not quite what we’re dealing with. While AI is software that’s capable of learning and improving, robots are simply a physical shell that acts and reacts in accordance with software that may or may not be intelligent and self-learning.
Artificial intelligence is a field of computer science research and a framework that focuses on simulating human intelligence.
Machine learning is an area of artificial intelligence that focuses on the ability of machines to learn from data.
Deep learning is an approach to machine learning inspired by the brain, and which follows decades of work on artificial neurons.
This article was excerpted from the C2 Montréal 2017 Minutes, which you can read in their entirety here.