How do you engage with your home and phone automation voice robots: Siri, Alexa, Cortona and their friends? I’ve seen a few different patterns, from timid adopters who self-consciously ask their phone to return a call, all the way to road warriors who dictate SMS as they make their daily commute and receive navigation advice in return. It’s a wide range, which is fine, because technology today is all about personal choice. Each person finds a level of engagement they are happy with – until two people in the same house both interact with the same robot, and their expectations and experiences are very different.
My wife Lyn and I have an inconsistent and troubled relationship with Siri.
I think it’s useful to consider why this is the case, because I think that in the future, our society is going to be much more reliant upon digital assistants to take on more and more important, urgent, and (potentially) dangerous tasks. I also think we will tend to take these assistants for granted. Cars and phones have changed our behaviour so that we’re no longer worried about how long it will take to get somewhere and exactly how to rendezvous when we arrive. In a similar way, our behaviour is adapting to fit around what our digital assistants can do for us.
I have a technical education and although convenience is a prime driver for the way I use Siri, I also like asking her to do little things – change the lights, choose some music, set an alarm, put the blinds up – just because I can. This is the future I’ve been promised ever since I watched The Jetsons after school in the 1960s and 70s – although I don’t live in outer space and Siri isn’t a robotic maid roving the house on wheels.
Siri is less physical, and yet almost omnipresent. I’m only just getting over the urge to show anyone visiting how it all works. “Hey Siri, set all lights to heliotrope,” I say, and the lights change to a light pink colour that no one knew had a name.
My spouse and soulmate, Lyn, has a music degree, and spent her professional life with opera companies and in human resourcing roles. She cares less about setting the lights correctly with a voice command, and more about trying to have Siri understand her request to play some Renaissance music. One of the women in my house knows who Arcangelo Corelli is and loves his Christmas Concerto; the other expects that every word we address to her will be in English, and while she can flawlessly find Jimmy Barnes or Lady Gaga, she doesn’t recognise the composer from the early 1700s. This generates endless frustration for me and Lyn until one of us finally keys in the name. Not the experience we were promised.
The difference, I think, is that being the artless technocrat I am, each time Siri does something correctly the first time I’m delighted and (re-)impressed at what is possible. Occasionally I’ll throw in a new way of phrasing a request, and I’ll smile gormlessly when Siri gets it right. Lyn rolls her eyes. Each time, I think about the programming and machine learning behind the miracle I’ve just seen performed.
Lyn, on the other hand, has a more prosaic view: of course Siri did that; it’s what you asked her to do. Lyn’s technological journey has been very different to mine. She cut her technological teeth on leading-edge smartphones and tablets, and as a result she has no possibility of delight; she has been trained to expect that the tech will understand what we want and always perform.
Unfortunately, my delight at Siri’s success is mirrored by Lyn’s disappointment and frustration when Siri says, “I’m sorry, I can’t do that,” or worse, turns on the wrong light, or puts on a rap artist from Suriname when Lyn wanted some Bach. So who is right and who is wrong? Am I too easily pleased, or is my life partner too harsh?
I think the answer is that she’s right: my wife, not Siri.
There is a difference between a gimmick and something that you can rely upon. While Siri still prompts joy in my heart, she clutters other peoples’ lives with irritation as she overpromises and underdelivers.
Until Siri gets better, people will continue to be disappointed. So when will this irritation end, and how will Siri and other home automation robots get better? And – beyond how to set the perfect music and lighting levels for a pearl anniversary dinner, when you didn’t google early enough to buy pearl earrings – what can we learn here?
Technology’s alignment to our needs has matured incredibly in recent years, to the extent that it now often falls short of our expectations, at least to begin with. More time is spent fitting the solution to the needs and wants and limitations of the people who will use a piece of technology. User experience and human-centred design are now at the heart of system design rather than add-ons. These learnings now need to be applied to the technology we use in our homes.
Fortunately, there has also been a change in how technologists understand what their users and customers want. We no longer publish a static system to our user community. Systems learn from what we, as users, feed them. They are pre-trained by scanning, listening and testing their responses against recordings of past behavioural data sets, and then those same learning circuits stay active when they are in the field. Siri, Alexa and Cortana also do this. They listen, they respond, they keep listening, and when we repeat our request, they learn.
Eventually, Siri gets it and recognises that “turn all lights off” requires the same action as “turn off all lights”, and that when we ask her to “play Bach” we aren’t asking for the last track again. So the answer is to keep trying, because you’re not just using a digital assistant: you (and everyone else using it) are simultaneously training it.
“Hey Siri, play Corelli, Concerto Grosso Adagio-Allegro” – one day, one day.