A friend once told me that for him, one of the weirdest moments as a parent was the realization that their kids have a relationship with each other, not just with him. That conversation always stuck with me, and when our own daughters started to have a bond with each other that’s completely separate from their bond with us, I understood what he meant. It’s just something that is, for some reason, very difficult to wrap one’s head around—these people you made, suddenly having lives apart from you. Today I’m thinking of that conversation again, but in a very different context—personal AI assistants, and what that means for how we design their interfaces.

But before I get there, let me take a step back and recap some of the recent conversations about personal AI. In The internet bundle is already here Dieter Bohn writes that AI personal assistants are a threat to net neutrality:

The bundle is already here, it came from places we haven’t been watching closely enough, and it has many names. There’s more than enough doomsaying about the issues related to Instant Articles, Internet.org, and Binge On. Instead, I’d like to take a minute to doomsay what could become the other opponents to the kind of free, transparent, and open internet we all want: Siri, Cortana, Alexa, Facebook M, and Google Now.

These intelligent assistants are great. I use them every day and expect I will continue to use them for, well, ever. But there’s a problem that’s built into them: they only seem to work with certain parts of the web and — here’s the real rub — certain apps.

Mark Wilson makes a similar argument in Why Every Gadget You Own Suddenly Wants To Talk To You, and then takes the argument further to imagine what happens when you have a bunch of non-neutral devices in your home:

But the problem with a scenario in which you can talk to anything is that you’re no longer talking to one thing. Only so many ears can live in one room. If I muse aloud that I need more shampoo in the shower, what hears me? Is it my iPhone sitting at the sink? Alexa networked in my apartment? Some new smart water nozzle from Kohler? […]

As consumers, we’re caught in the middle of the convenience. Do we choose to side with Siri, Alexa, or Cortana, and talk only to her, despite looming bias and the risk of growing dependent on a single voice—a voice that could take advantage of us? Or do we side with a free market that gives a voice to every stupid overzealous object in our lives, however confusing that may be, in a world where ordering milk becomes a bidding war on a commodities training floor?

Which future do you root for? They both sound horrible.

All of this is part of a big “the conversation is the interface” trend we’ve been seeing a lot of recently (see The search for the killer bot and 2016 will be the year of conversational commerce). From a design perspective the main challenge we seem to be thinking about is how to give these AI assistants the right personality (see The Next Phase Of UX: Designing Chatbot Personalities and The New Intimacy Economy). But I wonder if that’s the wrong AI design focus. I wonder if we should rather spend our time encouraging design for what Alan Cooper calls The Edges:

The difficulty in making these systems work smoothly comes from their edges, not from their centers. Each vendor builds a reliable and effective product, and through diligent testing assures that they meet high standards of performance. The only place where those standards fall is at the edges, where the maker is unsure of the requirements.

The edges are the interfaces with entities outside their control, outside their offices. Out there they are a little unsure of what they have to do and what forces affect them. Inside the company’s four walls they know exactly what they’re making, how it should behave, and what it should do. But for the entities outside those four walls, some measure of haziness creeps in, notably, the user.

Applied to AI assistants (and back to my parenting story), this means we need to start thinking about not just how humans interact with Siri and Alexa, but how Siri and Alexa interact with each other. There is, of course, a huge disconnect here between user needs and business goals. It would be very beneficial for users if different AI assistants could interact with each other, but that doesn’t help companies to strengthen their silos.

The trouble is that if we don’t figure out how to do this (and do it profitably), we might lose more than the battle of whose personal assistant wins. We might lose the war of personal AI getting any significant user adoption.

Michael Sacasas has an interesting viewpoint on the “machines are taking our jobs” argument. In Machines, Work, and the Value of People he argues that since we’ve so closely linked our value as human beings to the work we do, the issue of machines taking over hits us pretty hard:

So, to sum up: Some time ago, identity and a sense of self-worth got hitched to labor and productivity. Consequently, each new technological displacement of human work appears to those being displaced as an affront to the their dignity as human beings. Those advancing new technologies that displace human labor do so by demeaning existing work as below our humanity and promising more humane work as a consequence of technological change. While this is sometimes true–some work that human beings have been forced to perform has been inhuman–deployed as a universal truth, it is little more than rhetorical cover for a significantly more complex and ambivalent reality.

John Naughton wrote another interesting “the machines are coming for our jobs!” article1. This one is from the angle of “combinatorial innovation”—the idea that innovation happens when a bunch of disparate advances in technology come together in an unexpected way. His point in We are ignoring the new machine age at our peril is that it’s hard to see the implications of this kind of innovation:

The implications of [the self-driving] vehicle stretch far beyond the future of the automobile industry or even the future of transport. What it signals is that vast swaths of human activity – and employment – which were hitherto regarded as beyond the reach of “intelligent” machines may now be susceptible to automation. So we need to revise our assumptions about the future of work in the light of combinatorial innovation.

---

1. See, for example, The Machines are Coming and As Robots Grow Smarter, American Workers Struggle to Keep Up. ↩

In The Machines Are Coming, Zeynep Tufekci talks about the kind of tasks that are being automated by machines:

Today, machines can process regular spoken language and not only recognize human faces, but also read their expressions. They can classify personality types, and have started being able to carry out conversations with appropriate emotional tenor.

Machines are getting better than humans at figuring out who to hire, who’s in a mood to pay a little more for that sweater, and who needs a coupon to nudge them toward a sale. In applications around the world, software is being used to predict whether people are lying, how they feel and whom they’ll vote for.

This is not a new topic. Back in 2012, Kevin Kelly proclaimed in Better Than Human: Why Robots Will — And Must — Take Our Jobs:

It may be hard to believe, but before the end of this century, 70 percent of today’s occupations will likewise be replaced by automation.

At the end of last year Claire Cain Miller wrote for the New York Times that As Robots Grow Smarter, American Workers Struggle to Keep Up:

Although fears that technology will displace jobs are at least as old as the Luddites, there are signs that this time may really be different. The technological breakthroughs of recent years — allowing machines to mimic the human mind — are enabling machines to do knowledge jobs and service jobs, in addition to factory and clerical work.

Who knows if this fear is going to turn into reality or not — there are lots of counter-arguments as well (For example, Nicholas Carr has a really interesting historical perspective in Should the Laborer Fear Machines?).

Still, I find the discussion fascinating — especially as it relates to the balance of power in workplaces. Tufekci continues:

Machines aren’t used because they perform some tasks that much better than humans, but because, in many cases, they do a “good enough” job while also being cheaper, more predictable and easier to control than quirky, pesky humans. Technology in the workplace is as much about power and control as it is about productivity and efficiency. […]

This is the way technology is being used in many workplaces: to reduce the power of humans, and employers’ dependency on them, whether by replacing, displacing or surveilling them.

Maybe that’s the real cause for concern here. Not that jobs might go away (although that’s certainly worrisome too), but that power will continue to shift to employers and away from employees.

Benedict Evans wrote a characteristically brilliant analysis in What does Google need on mobile? Here’s a taste of his conclusion about Google’s challenge going forward:

The key change in all of this, I think, is that Google has gone from a world of almost perfect clarity—a text search box, a web-link index, a middle-class family’s home—to one of perfect complexity—every possible kind of user, device, access and data type. It’s gone from a firehose to a rain storm. But on the other hand, no-one knows water like Google. No-one else has the same lead in building understanding of how to deal with this. Hence, I think, one should think of every app, service, drive and platform from Google not so much as channels that might conflict but as varying end-points to a unified underlying strategy, which one might characterize as ‘know a lot about how to know a lot’.

Don’t miss this article, the whole thing is great.

Tim Harford has an excellent critique of the statistical issues with the “big data” trend in Big data: are we making a big mistake? First, there’s this:

But the “big data” that interests many companies is what we might call “found data”, the digital exhaust of web searches, credit card payments and mobiles pinging the nearest phone mast.

I still love the term “digital exhaust”. I first saw Frank Chimero use it in the context of social media when he said (in a post that’s now gone from the internet):

The less engaged I become with social media, the more it begins to feel like huffing the exhaust of other people’s digital lives.

But back to big data. The big problem (see what I did there?) is that statistical problems don’t just go away when you have more data. In fact, they get worse. For example:

Because found data sets are so messy, it can be hard to figure out what biases lurk inside them – and because they are so large, some analysts seem to have decided the sampling problem isn’t worth worrying about. It is.

The article goes into the detail on this, and I think it’s important for us to recognize the limitations of big data before jumping on the bandwagon.

In How do Smartphones Affect Human Thought? Jenny Davis addresses the recent research behind the “Smartphones are Making Us Stupid” narrative:

[The research] hypothesis implies (though does not state) a research question: How does smartphone usage affect cognitive processes? This is an important question, but one the research was never prepared to answer thoughtfully. Rather, the authors recast this question as a prediction, embedded in a host of assumptions which privilege unmediated thought.

This approach is inherently flawed. It defines cognitive functioning (incorrectly) as a raw internal process, untouched by technology in its purest state. This approach pits the brain against the device, as though tools are foreign intruders upon the natural body. This is simply not the case. Humans [sic] defining characteristic is our need for tools. Our brains literally developed with and through technology. This continues to be true. Brains are highly plastic, and new technologies change how cognition works. Our thought processes are, and always have been, mediated.

The response echoes many of the points Clive Thompson brings up in Smarter Than You Think (my review) — namely that technology can be great augmentations to human thought. It’s not all bad.

Kashmir Hill asks an interesting question: Who do we blame when a robot threatens to kill people?

Last week, police showed up at the home of Amsterdam Web developer Jeffry van der Goot because a Twitter account under van der Goot’s control had tweeted, according to the Guardian, “I seriously want to kill people.” But the menacing tweet wasn’t written by van der Goot; it was written by a robot.

He goes on:

Bots will be bots. They won’t know if they’re doing something wrong unless we program them to realize it, and it’s impossible to program them to recognize all possible wrong and illegal behavior. So we’ve got challenges ahead. In the short term, [Clément Hertling, a Paris-based university student who wrote the software that powered the bot] suggested Twitter — and any other platforms bots might live on — could solve the offensive speech problem by allowing bots to self-identify in an obvious way as bots. “That would allow people (law enforcement included) to ignore what they say when it becomes problematic.”

This issue only gets scarier as the question expands to wondering what happens when we put self-driving cars in morally ambiguous situations.

In The Cathedral of Computation Ian Bogost makes the argument that algorithms have replaced religion for many people:

Here’s an exercise: The next time you hear someone talking about algorithms, replace the term with “God” and ask yourself if the meaning changes. Our supposedly algorithmic culture is not a material phenomenon so much as a devotional one, a supplication made to the computers people have allowed to replace gods in their minds, even as they simultaneously claim that science has made us impervious to religion.

It’s a long article but very much worth reading, especially for the conclusion:

Algorithms aren’t gods. We need not believe that they rule the world in order to admit that they influence it, sometimes profoundly. Let’s bring algorithms down to earth again. Let’s keep the computer around without fetishizing it, without bowing down to it or shrugging away its inevitable power over us, without melting everything down into it as a new name for fate. I don’t want an algorithmic culture, especially if that phrase just euphemizes a corporate, computational theocracy.

But a culture with computers in it? That might be all right.