A lot of my conversations at CES this year have been with ambitious companies who have a product or service idea they want to present at next year’s event. And there's certainly time for that – our product realisation team are used to the “on the shelves by Christmas” conversation at this time of year.
I expect many new products and services at CES 2019 will include an machine learning, AI or deep learning element. AI will touch every business over the next few years, whether or not they're tech companies, which is why we have invested in technology to help our clients rapidly take advantage of AI.
The power of AI is not to replace humans, but to enhance what we're able to do, and to improve the decisions we make, even with very sparse data – AI is good at filling in the gaps. We’ve developed the concept of “explainable AI”, to provide insight into how an AI is making a decision. This is critical for situations where, for example, an AI is helping diagnose or monitor a condition, or for a safety critical system, but it’s generally good principle to show you what an AI is doing with your data rather than it being a mysterious black box. The Next Web recently tried out our My Accent demonstration and had fun trying to fool it.
AIs can also be deployed to do human-like tasks that you simply wouldn’t hire people to do, and certainly not multiple people. For example, Vincent uses seven neural nets forming a team (a type of Generative Adversarial Network) that works together to turn sketches into art in real time. It’s very hard to imagine anyone hiring a team of artists to do this. Yet we had people on the stand using Vincent as a genuinely creative tool – to interact with the AI and be influenced by what the AI was creating as well as their own creativity.
So, if CES 2018 was a year of promise for AI, I think 2019 will be the year it becomes mainstream.