“Back in 2008, there were people, including me, who predicted the current trouble with Facebook, Now we can’t even solve those problems any more, really: two billion people use Facebook and Facebook apps like Instagram and WhatsApp. What can we do about it now? We need to predict problems and solve them before they arise.
“At present, there are about fifty million people using virtual home assistants like Alexa and Google in the United States alone. Their popularity has grown very rapidly, when you consider how new they are. The more they understand their users and the more electronic devices can to work together, the more popular they will be.
“They are the gate and they have the platform. That makes them dangerous in three ways. They decide what you see, they collect lots and lots of personal data and once they have that data, they learn from it and try to improve. Let’s say virtual assistants could sends a patient’s blood pressure reading to the doctor; if enough doctors monitor blood pressure that way, the artificial intelligence can collect a huge amount of data. It can find patterns in that data and attempt to improve blood pressure levels, meaning that device assumes the doctor’s role. We’ll start to see such changes everywhere.
“We’ll be able to measure people very, very accurately. The main question is: will there only be one or two companies who profit from it, or will there be more? Even if you trust those companies with your data, it’s better for innovation if there is open competition.
“At Stanford, my colleagues and I are working on Almond, an open-source virtual assistant: everyone is allowed to see the code, everyone is allowed to use it, and everyone is allowed to build applications or things that work with Almond. You decide which data it shares with whom and you can program it yourself.
“At the moment, I use it mostly to remind me to charge my car, but you can get it to do all sorts. One integrated messenger, instead of lots of individual ones. Commands like: ‘If the price of Bitcoins drops below two thousand dollars, send a message to my colleagues via Slack’ or ‘When I leave the office, turn on the heating at home.’
“Let’s say you have asthma. You can blow into a lung monitor connected to your assistant. You can keep track of any changes so when your doctor asks, ‘Send me an email if the volume drops below such-and-such’, you give your assistant permission to send that information. It could warn you if the air is too polluted to go running and warn your parents if you are admitted to hospital.
“You can program it in natural language. Artificial intelligences are now so far advanced they can make the transition from a graphic user interface like Windows or your phone’s touchscreen to the linguistic user interface. You give it simple ‘if this happens, then do that’ commands and it generates the corresponding code. It works because people understand that they are talking to a computer. You don’t say ‘make sure the room temperature is in my comfort zone’, you say ‘turn the heating up to 21 degrees’.
“We’re putting more and more new things in Almond, even though the basis is not good enough yet. I simply don’t have the people to do it. I want to set up an alliance with the industry and hopefully they can provide the manpower for it. Why would they want to? Because the rise of virtual assistants is at risk of becoming a duopoly. The market for home assistants seems to be split in two: at electronics fairs, you see products that can either work with Google Home or Amazon’s Alexa, or both, and nothing else really.
“It’s the same situation as early on in the networking world. Everything used Cisco and if you wanted to change something, you had to sell your company to them. A colleague of mine at Stanford developed the so-called Software Defined Networking and all sorts of businesses, including Google, collaborated because everyone can innovate in an open market. It also helps that the assistants are a new platform. The market for operating systems was dominated by Windows – open Unix/Linux never really caught on. But then smartphones arrived on the scene and they operated on Android-based Linux. The Windows Phone lost out to it.
“I’m counting on there being enough companies who want to compete with Amazon and Google. There’s Apple, for starters, and large chain stores like Safeway. After all, Alexa won’t make it easy for you to buy things from one of Amazon’s rivals. Together, the other parties are quite powerful, and who wants to be under the thumb of those two companies?”
Monica Lam is Professor of Computer Technology at Stanford University. She was in Leiden for the Ada Lovelace Distinguished Lectures, for which LIACS, Leiden’s Computer Science institute, invites female speakers. Professor Lam is developing virtual assistants that can be programmed in ordinary human language. Moreover, she is an expert on compilers (which translate codes from one programming language to another) for supercomputers and web application security