The world’s 230 million knowledge workers are frazzled. Modern life is an interminable cacophony of emails, notifications, messages, alerts, feeds, data and information. 70% of us look at our phones within 30 minutes of waking up. All this causes stress. With multiple notifications on multiple apps on multiple pages of our devices, where do we start? Who will help us?
Fortunately, almost all of us already have a personal assistant. It’s a piece of software on a device you own: the intelligent assistant (IA). We carry IAs around on our laptops (Microsoft’s Cortana), phones (Google Assistant, Apple’s Siri, Samsung’s Bixby) and smart speakers (Amazon’s Alexa, Baidu’s Little Fish). You probably have more than one. There are an estimated one billion IA-enabled devices in the world today. With smartphone penetration in the UK and US approaching 70%, it’s easy to believe that there will be as many intelligent assistants as human beings in a just few years. That billions of people will soon have an assistant 24/7 is a staggering prospect.
Driving this change, of course, are the tech giants. IAs are one of their top priorities. Google Assistant stunned the crowd at I/O 2018 with natural-sounding phone calls to unassuming restaurant and hair salon staff. Siri is now making recommendations based on your activity and schedule. Alexa is starting to understand general user intent. Google’s big advertising slogan on digital billboards today is “Make Google do it.”
Yet take-up so far is low. The game has not changed as many of us predicted it would. We still only spend a few minutes a month engaging with our IAs. The idea that this software will genuinely save us time and increase our productivity is not yet taken seriously by business people or the general public. Why aren’t IAs working better and harder for us? Let’s start with what they can actually do.
Today, we instruct our IAs on a narrow range of tasks. But they can already do a lot more than this. Though different manufacturers describe their products differently, there’s a substantial commonality:
- Change settings: “Turn on airplane mode,” or “Find my phone.”
- Plan your day: “Remind me to send that invoice,” or “Set up a meeting at 9 a.m.”
- Find information: “Who won the Word Cup?” or “Translate ‘me gustaría entender’”
- Perform tasks: “Make a dinner reservation at Woods Hill Table,” or “Reorder tortilla chips.”
- Entertain ourselves: “Play OK Computer by Radiohead” or “Show photos from yesterday.”
- Communicate: “Call Andrew” or “Tell Susan I’ll be right there.”
Their extensive capability derives from the vast app ecosystem (6.2 million apps). IAs leverage apps (or “skills” in Alexa’s terminology) to do what they do. The IA takes a request, decides which of the apps available can deal with it and how to use the chosen app to do so. At its simplest, an IA will merely launch the app (“Siri, launch Photos”). Voice-activated commands start to add value when used more specifically, i.e. with a “deep link”: “Show photos from last May” or “Text Susan I’m running late.” In other words, IAs use app functionality as the building blocks of their administrative know-how.
This is why IAs are so powerful even today and why we should summon the genie from the bottle more than we do. The app ecosystem is as vast and rich as it is precisely because the apps there cater to such a wide variety of human needs and desires. IAs leverage this enormous resource. The list of apps on your phone is another way of describing the abilities of your IA; its resumé, if you will.
So why aren’t we using them?
To date, the hardware and software has not quite been up to the task. Siri has been around since iPhone 3 (2009) but it’s never quite cut it as a credible time saver or productivity enhancer. Nine years on, more apps, better apps, a coordinated app ecosystem and ever-increasing processing power have addressed many of these performance deficiencies. The product is well beyond viable now.
So the bottleneck lies with us, the everyday users of these things.
The power of your IA is a function of how many useful skills it has and how much you invoke them. To broaden the skillset, be aware of the full virtues of pre-installed apps and choose additional apps proactively and strategically. To invoke them more we need to work both on our comfort with and our knowledge of IAs; our symbiotic relationship with machines requires us to adapt as well. We need to develop a thicker skin to feel less self-conscious talking to non-human agents.
For some, privacy and data has put them off using their IA. Cambridge Analytica and GDPR are massive privacy stories from this year which will have exacerbated such concerns. The tension between convenience and privacy is felt by us all, consciously or subconsciously. Mary Meeker describes this as a “privacy paradox”: internet companies use data to make low-priced products better so users spend more time on them which leads to both more compelling products and increased regulation. But if the ultimate arbiters are consumers, rather than regulators, it seems that convenience will ultimately triumph. We buy 10 million smart speakers every quarter and these devices listen to every word we utter in the place we live, by definition. We’re rapidly getting comfortable with the idea.
We also need to understand what IAs can do for us and, to that end, to find a way to think about IA capability. Well, here’s a way to think about it: IAs help us use our favorite hardware (smartphones, laptops, smart speakers) more efficiently by launching apps at the point and in the way that you need them. Since we’ve all become so comfortable using apps, thinking of IAs in this way should make the prospect seem less daunting, more within reach.
IAs will do a lot more very soon.
Today’s mostly reactive IAs will be more proactive. We’ve seen glimpses of this already. An IA will do this by combining data it has on you across different domains to make suggestions. For example, the IA will see that it’s your husband’s birthday (extracting this from your calendar) and suggest you call him, notice where you are in a new city and recommend some good restaurants nearby (along with directions to get there). This fall, Siri will be able to bundle up actions from different apps (called Shortcuts) so, for example, you might tell Siri you’re “Heading home” and it will then offer you directions, text your family you’re coming, set a thermostat and summarize all this with an ETA.
Less trivially, IAs will be able to help with our physical, emotional, and mental well-being. They will take the right combination of cues — heart rate, spending patterns, location, social media usage, smartphone usage, what we say and write — to make useful recommendations, sensitively. IAs will also help lift individual and collective human capability by serving up relevant recommendations for learning, information and knowledge, just when you need them, just where you are. Recommendations can then be taken or not; there’s a final confirmatory step which helps us to maintain a sense of control and agency.
The Duplex demo went further than recommendations. Instead of the IA just thinking to tee up a call with the right restaurant for the human to take over, it went right ahead and had the whole conversation. That’s partly why some people felt uncomfortable with it. Google has carried on at full pelt, astounding journalists with live demos of Duplex a couple of weeks ago. These advances force us to face some important questions. What is admin? What class of tasks are beneath us? How do we strike the right balance between efficiency and control? As we carve out the shape of artificial intelligence, we can’t help but reshape what we mean by human intelligence and humanity.
IAs will significantly change human behavior in the coming years. As a forward-thinking professional, be aware of it, understand it, anticipate it, and experiment with it. You will do less but achieve more. You will also cultivate a greater sense of the benefits of artificial intelligence, from natural language processing to feedback mechanisms and machine learning. You will feel better assisted, and less frazzled.