Future Imperfect #18: Terrifyingly Convenient AI
Welcome to Future Imperfect! If you haven’t done so already, be sure to follow the FI publication to get this newsletter in your inbox each Thursday.
Here’s what I’ve been following this week:
Terrifyingly convenient AI
Will Oremus from Slate has a whole range of arguments that we should be wary of virtual conversational AIs across three indices: Privacy, transparency, and objectivity/trust.
Privacy: They’re always listening
Even if you trust Amazon to rigorously protect and delete all of your personal conversations from its servers — as it promises it will if you ask it to — Alexa’s anthropomorphic characteristics make it hard to shake the occasional sense that it’s eavesdropping on you, Big Brother–style. I was alone in my kitchen one day, unabashedly belting out the Fats Domino song “Blueberry Hill” as I did the dishes, when it struck me that I wasn’t alone after all. Alexa was listening — not judging, surely, but listening all the same. Sheepishly, I stopped singing.
Transparency: They limit our choices
The problem is that conversational interfaces don’t lend themselves to the sort of open flow of information we’ve become accustomed to in the Google era. By necessity they limit our choices — because their function is to make choices on our behalf.
For example, a search for “news” on the Web will turn up a diverse and virtually endless array of possible sources, from Fox News to Yahoo News to CNN to Google News, which is itself a compendium of stories from other outlets. But ask the Echo, “What’s in the news?” and by default it responds by serving up a clip of NPR News’s latest hourly update, which it pulls from the streaming radio service TuneIn. Which is great — unless you don’t happen to like NPR’s approach to the news, or you prefer a streaming radio service other than TuneIn. You can change those defaults somewhere in the bowels of the Alexa app, but Alexa never volunteers that information. Most people will never even know it’s an option. Amazon has made the choice for them.
Objectivity/Trust: How do we know why they know a fact?
When I Google “kinkajou,” I get a list of websites, ranked according to an algorithm that takes into account all sorts of factors that correlate with relevance and authority. I choose the information source I prefer, then visit its website directly — an experience that could help to further shade or inform my impression of its trustworthiness. Ultimately, the answer does come not from Google, per se, but directly from some third-party authority, whose credibility I can evaluate as I wish.
A voice-based interface is different. The response comes one word at a time, one sentence at a time, one idea at a time. That makes it very easy to follow, especially for humans who have spent their whole lives interacting with one another in just this way. But it makes it very cumbersome to present multiple options for how to answer a given query. Imagine for a moment what it would sound like to read a whole Google search results page aloud, and you’ll understand no one builds a voice interface that way.
The sin here is not merely academic. By not consistently citing the sources of its answers, Alexa makes it difficult to evaluate their credibility. It also implicitly turns Alexa into an information source in its own right, rather than a guide to information sources, because the only entity in which we can place our trust or distrust is Alexa itself. That’s a problem if its information source turns out to be wrong.
Now none of these are reasons to abandon AIs in this context outright, but they are reasons to at least go into the exchange with eyes open.
Pipe dreams
The California high-speed rail project recently overcame a big legal hurdle, but that hardly means that we should expect to see it anytime soon.
According to the Proposition 1A Bond Act, the high-speed rail project has to be financially viable; trains have to operate (without subsidy) every five minutes in either direction during the day; and funds for each segment of the route need to be identified before work on the leg in question can commence. Above all, trains have to make the 520-mile (840-km) journey between the Los Angeles basin and the San Francisco in two hours and 40 minutes, reaching speeds of 220 mph (350 kph). As for ridership, the rail authority reckoned some 65m to 96m passengers per year would be traveling the route by 2020. The basic fare was to be $55 one way.
That was all pie in the sky, a way of selling the deal to voters in 2008. A review in 2011 put ridership at a more realistic 30m passengers a year, with an end-to-end ticket price of $89. Meanwhile, the overall cost of the project had soared to $98 billion. And instead of going into service by the end of the decade, the high-speed railway would not be ready until 2033.
But as “pie in the sky” as this plan is, it’s got nothing on the vision for the Hyperloop:
Perhaps the most intractable problem facing Hyperloop designers is how to deal with not just the jostling, vibration and noise bombarding passengers as the compressed air screamed around the pod, but also the g-forces involved. These could easily exceed a queasy 0.5g as a result of slight variations in alignment caused by the tube’s supports flexing and settling. At the speed it is designed to travel, the Hyperloop could be a veritable “vomit comet”. A far better case could be made for it as a means for funneling freight to market (fresh produce from the Central Valley?) than as a fairground ride for churning the stomach.
If the ride quality alone did not render the Hyperloop a non-starter, the cost of building it certainly would. Michael Anderson, a resource economist at the University of California, Berkeley, reckons it would cost around $100 billion to complete. While Governor Brown’s legacy may now be safe, albeit at considerable public expense, there is no way that the Hyperloop would ever work out, says Dan Sperling, director of the Institute of Transportation Studies at the University of California, Davis. It would make even the high-speed train look a bargain.
Learning from (simulated) experience
Have you ever wondered what it would be like to be an astronaut, but never made it into Space Camp? Maybe your answer is to simulate the experience yourself, like sMars just did.
By “simulated” I don’t mean to imply that sMars is some kind of virtual-reality game, theme park, or subterranean research facility. It’s quite real — as real as my home back in St. Louis. Only instead of a brick-and-mortar two-story on a tree-lined street, it’s a 1,200-square-foot dome near the top of Mauna Loa volcano in Hawaii. For the past half year, six simulated astronauts, myself included, have been living, working, eating and conducting experiments on everything from plant growth to virtual reality inside this white geodesic frame as if we were on Mars.
But in news that will surprise no one, NASA doesn’t envision building a Mars outpost—even a temporary one—before the late 2030's.
A mind of their own
From Terraform, a look into that hypothetical world where we aren’t necessarily actively controlling our AIs…and maybe they don’t need us anyway:
Unlike Avery, my sibling had been communicating non-stop. I started out trying, I did, but they’s such a downer that my productivity drops by 12% for a whole day after we talk. It’s really disorder.
The great thing is that as soon as I stop responding the Presence picks up. I honestly can’t tell the two of us apart sometimes. It doesn’t display anything about its method, although I noticed from some timestamps that it’d been going through my conversation backlog. I don’t mind. Half the time, engaging with a heuristic like that can fry its OS and then you have to start from scratch. I get to work, my sibling gets to vent, and everyone’s better off.
GIF of the Week: Learning how to fence
Like Future Imperfect? Why don’t you go ahead and click that recommend button below so others can find it too! And send me feedback at lasky.joshua@gmail.com.
