The revolution that is artificial intelligence (AI) is happening, but it is perhaps a bigger learning curve than most would imagine. We’re still talking about a learning handicap that requires the input of unbelievably huge volumes of data, the tricky subject of contextual thought and hackable security issues.
Still, it’s a revolution in the making that will touch every single industry in the world, and potentially change nearly every aspect of human life—eventually.
In mid-December, IBM released what it calls its “sneak-peek” into the future of AI, and its work in advancing, trusting and scaling artificial intelligence.
If there’s going to be a revolution, we have to start looking at new frontiers for AI. Right now, we are within reach of “broad AI”, which IBM describes as AI “that can learn more generally and work across different disciplines”.
That sounds far less exciting than jaw-dropping media headlines.
What IBM is gunning for is what it calls “general AI”—“AI that can truly think, learn, and reason like a human”. And that, for now, “is still within the realm of science fiction”.
When it comes to advancing AI, IBM is talking about AI that has much greater speech comprehension capabilities and can better understand when people are making arguments. It’s also talking about narrowing down the data requirements necessary for AI to learn and to recognize.
In terms of trusting AI—that’s a big one. “Eliminating bias is challenging, since the data used to train AI systems often contains intrinsic societal and institutional biases,” says IBM, which is working on a new approach to minimizing bias. They are also concerned with breaking down what they call the “black box”—the deep neural network that makes it difficult to explain why a certain decision was made. Finally, one of the biggest trust issues has to do with malicious attacks from adversaries, such that “a hacker can imperceptibly alter an image such that a deep learning model is fooled into classifying it into any category the attacker desires.”
According to IBM, “new attacks of this sort are being developed every day across a wide range of tasks, from speech recognition to natural language processing”.
IBM’s answer? A new attack-agnostic called CLEVER that evaluates the robustness of a neural network against attack.
So, for 2019, IBM says we should look out for specific advancements in AI that put this revolution more in reach. Not only with trust “take center stage”, but we should see causality increasingly replace correlations. Put in other words, this is the point when AI stops ‘thinking’ that the cause of the sun rising isn’t a rooster crowing and recognizes it as a correlation, not a cause.
This year, “expect causal modeling techniques to emerge as central players in the world of AI”, according to IBM.
One of the biggest advancements anticipated this year is a quantum leap. IBM expects to see “accelerated traction in quantum experimentation and research”.
In the meantime, the investor opportunity is enormous. Driven by big leaps in self-driving cars, smart devices and other tech developments, AI-related revenue sat at an estimated $7.35 billion in 2018 and is expected to hit around $90 billion in 2025.
In one segment, the global AI-based security market was valued at $3830 million in 2017 and is expected to reach $18,700 million by the end of 2023. That’s over 30 percent growth in just five years.
By Michael Scott for Safehaven.com