Amazon Echo is a widely acclaimed, voice controlled AI device. It is built on the technology of a relatively unknown British company called Evi. Founded in 2005 by entrepreneur William Tunstall-Pedoe, the company was built to develop a software that accurately interpreted questions and framed more natural, conversational answers. After a successful seven years the technology crossed the radar of Jeff Bezos’s Amazon, who quickly snapped up the software in 2012 at a pretty price.
When it hit the market in 2012, the technology at the time was called Evi 11, and was positioned to compete with Apple’s Siri. However this had not been the vision of Tunstall-Pedoe, who set out to build the raw technology that had a signifiant impact on the world. 11 years after its initial inception, he can now celebrate Evi’s real world impact. He says “these technologies are now good enough that they are able to create useful products that change lives and are used daily”.
Before Evi’s invention, Tunstall-Pedoe, who lives in Cambridge, built his reputation as an AI mastermind, and was known for programming computers to crack cryptic word puzzles. He also developed the Anagram Genius software that uses AI to turn words into anagrams. Dan Brown used that software to devise the anagrams in The Dav Vinci Code. After three years at Amazon, Tunstall-Pedoe left to pursue other AI projects. He recently spoke to Wired Magazine about where voice recognition is going, and what he’ll look to do next. We have included the excerpts from the interview here:
Before building Alexa, the AI that underpins Echo, you built an AI that could solve cryptic crosswords. How did this lead to Evi?
It sounds trivial in comparison, for the size of the problem, but solving cryptic crosswords is a classic AI problem in many ways. It's something computers find very hard, and it requires intelligence. In terms of world impact it's obviously very small, but I was very proud of the technology2. The origins of Evi are about a desire to apply these technical skills and understanding to much bigger problems that can affect the lives of billions of people. For instance, the way we operate computers, with buttons or custom interfaces or guessing keywords, for me is not the way that computers are going to work in the future. Surely the most natural way to operate a computer is just to ask it for what you want? That vision is what drove me to found Evi.
What is it that makes Evi competitive as a technology?
One of the big unsolved problems in AI is the ability to understand natural language. The reason search engines still largely work with keyword search, statistics and snippets of text is that the technology doesn't understand what's in a document. There's no deep understanding that comes from reading a document. We haven't solved that problem, but the knowledge that powers the Evi platform is a knowledge base of structured data, including common-sense knowledge, that's in a form computers can understand3. So it's not going to a collection of documents - it's not like a search engine.
The other thing that's pretty unique is its ability to reason with knowledge. So we can take a question that has never been asked before, find multiple facts in the knowledge base and chain them together, combining them to create new knowledge that's needed to answer. Our ability to exploit that knowledge base is where the power comes from. This results in many more of the user's questions being answerable than would otherwise be the case.
How do you predict this will change our reality?
In ten years, people will expect all technology to respond to voice. In every building and vehicle, a computer system will respond to spoken requests and control the technology. Light switches and other controls that we see today may still be present, but will just be the manual alternative. People will also take for granted instant access to all their private data and all human knowledge, just by asking. It's coming sooner than people might realise.
As our idea of intelligence changes and refines, surely the definition of AI will evolve too?
Basically, there is one object in the Universe that everybody acknowledges is intelligent, and that's the human brain. So we have one reference point for what intelligence is. Computers are built in a completely different way to brains, though there is quite a bit of work now trying to merge cognitive science and AI, looking at how neurons work and taking inspiration from that. A lot of the recent big advances in artificial intelligence, such as those in computer vision, have come from what's known as deep neural networks, which are inspired by the way the brain works.
So, mimicking the human brain should be AI's ultimate goal?
To be clear, there are things that computers can do way better than the brain can do. Nobody remotely challenges a computer's ability to do arithmetic better than a human. So it's not so much drawing level with the brain. It's about creating computer systems that surpass what people can do, for the benefit of people.
What's next for you?
AI is pushing the boundaries of what's possible with computers. I'm looking for the next really big thing, that will positively impact billions of people. That's where my focus is. I am keeping an open mind for exactly what that will be. I hope to settle into something new in a few months' time.