I don’t want to chat with the command line, just make Siri awesome.
Apple has one of the most popular messaging services around, iMessage. They have an AI “bot” already used daily by millions across the globe, Siri. They have an extensive and robust, if imperfect, marketplace for third-parties, the App Store. If Apple can integrate these three services: messaging, a digital assistant, and a third-party marketplace, then I think they’ve got a great shot at dominating the wave of bots that is supposedly coming.
Apps Aren’t Broken
Most people aren’t going to find the process of discovering and engaging bots any easier than discovering and installing apps. Is the App Store a great place to discover apps? Not really. But the whole world knows how to install an app if you tell them its name.1 Bots per se don’t make discovery any easier. There’s still a marketplace or directory, and pressing “start chat” versus “get” really doesn’t change the process in any meaningful way. If anything, a separate “bot store” within an app (say, Facebook’s Messenger) is a more confusing arrangement for most folks.2
Bots aren’t always going to be better than apps. Bots have the potential to be really great when a person has a good idea of what they want, but I’m not convinced they’re better when someone is exploring. For example, if I want to make reservations at the Turtle Club restaurant here in Naples, I can see the appeal of telling a bot to “make a reservation for two at the Turtle Club tonight”. However, if I’m looking for a romantic restaurant the Foursquare app is going to be much better and faster at showing me all my options and detailed information than a chat-style interaction even if that chat is enhanced beyond just plain text.
As far as interaction goes, I’m not sure that typing out “make a reservation for two at the Turtle Club tonight” is really a lot easier than the few taps it’d take for me to do the same in the OpenTable app. As anyone who’s ever used the command line knows, an important difference in the experience of typing a command and using an app is that you don’t know if your command is properly interpreted until after you’ve made it.3 In an app, you get confirmation and guidance as you go. This isn’t an insurmountable issue; Flexibits’ Fantastical does an outstanding job in this regard, taking natural language input and visually confirming to the person typing how it will be interpreted when they press enter. Similarly, the implementation of bots in Telegram thus far seems to recognize this issue and tries to address it by showing live results for @gif or @YouTube searches. Still, I don’t think long complicated commands being typed into computers is really the future. It sounds a lot like the past.
But you know what sounds a lot like the future?
Talking to computers.
The most natural evolution, it seems to me, would be for Apple to allow third-party apps (not just bots by another mechanism) to extend Siri. Of course, this idea isn’t new, since the introduction of Siri third-parties have wanted Apple to provide an API for extending her capabilities, or “skills”.4 What’s new is the renewed focus on natural language interaction that the current bot trend has encouraged. There is also now a precedent for Apple allowing third-party apps to extend and change system-wide services on iOS. (C.f. Safari ad blockers, spotlight search results, or even today widgets.) When I imagine talking to bots, or at least starting the conversation by speaking, this really starts to sound exciting.
Especially with Siri on Apple Watch.
Ask people who love Apple Watch why they do, and you’re probably going to hear these things again and again: Health & Fitness, Notifications, Complications, and Siri.
The convenience of “Hey Siri…” on my wrist, even though there are very few use cases, still surprises me. Every morning I ask Siri to set a timer while my coffee steeps. Every afternoon when I go on a walk I ask her to “start an outdoor walk with no goal”. When I’m trying to focus on a problem or develop an idea and some mundane to-do or distraction pops into my head, I just raise my wrist and ask Siri to remind me later. When I get in the car and need directions, I raise my wrist and ask Siri. When I’m cooking and need a quick conversion, or calculation, I raise my wrist and ask Siri.
Allowing third-parties to extend Siri would obviously make the service more useful by increasing the number of actions available to Siri. However, as others have suggested, also creating a dedicated iMessage thread for more advanced interactions with Siri sounds like a great idea. As it is today, Siri on iOS 9 does display rich “widgets” in response to many requests such as setting timers, showing sports scores, requesting definitions, peeking at Wikipedia entries, and on Apple Watch you can even ask Siri to display glances (including ones you haven’t set to display when you swipe up) and she does so inline.5 Something like glances or today widgets could be part of a Siri API and allow many of the more interesting bot interactions that folks have imagined or built on other platforms.
Being able to access my chat history with Siri, as well as to interact without speaking sometimes, is an attractive prospect too. It is occasionally frustrating that the Siri interface on iOS and Apple Watch just disappears and there’s no history. Sometimes I want to switch between a Siri screen, such as a conversion of ounces to tablespoons, and my favorite recipe manager, Paprika. Or I want to copy and paste the response to a query elsewhere. These and other benefits to a chat-style interface with a “bot” are clear, I think.6
But I don’t want 50 “bots” junking up my messages with actual humans with whom I’m having conversations.
Email already suffers from too many impersonal, automatic, messages crowding out actual human-to-human conversation. Avoiding email and its legacy distractions is probably no small part of why iMessage, Slack, Telegram & the like are so attractive to begin with. Importing all that junk into their systems seems like a bad long-term decision. As cool as asking Siri to book a dinner reservation for me sounds, I would hate to have an iMessage thread with the OpenTable bot, and a hundred others, mixed in with my conversations with friends, family and business folks. By implementing dedicated Siri iMessage thread, or something similar, you’d have a truly unified interface for chatting with a “bot” without overwhelming the messaging app with tons of different bots.
I find today’s bots to be kind of boring. Cool, sure… I guess. But boring. I don’t expect to see any welcomed to iOS by Apple. I expect instead to see Apple allow third parties to extend and enhance Siri through apps, taking all those things which seem attractive about bots (most especially the natural language interaction) but in the far more convenient and accessible form of Siri, on your wrist or desk, and in your pocket or living room. I also hope that Apple will introduce some kind of iMessage-style method of interacting with Siri and offer developers a way to include interactive widgets in Siri responses. Finally, perhaps we’ll eventually see some of Notification Center folded into Siri as simply our digital assistant’s way of proactively reaching out to us with a message.
This Jarvis-like scenario for Siri sounds more magical to me today than chatting with the command line. If there’s anything Apple does well, it’s making today’s magical, tomorrow’s mundane.
Update 2016-04-18 at 2:50 PM: Just saw MacStories post that WWDC ‘16 was “announced” via Siri. Perhaps it was a leak more than an announcement, but having published this a few hours before, I’m obviously expecting great things from Siri and wouldn’t be surprised if this was done with purpose. (I added the featured image to this post because it’s just perfect now.)
Sure, if you search “Twitter” you might not find Tweetbot but if you’re looking for Tweetbot you’ll get it. ↩
As others have pointed out. Also, this seems like a really unstable position for Facebook. Would buying a Kindle book through a chat with Amazon-bot in Messenger be subject to a fee from Apple? Would Facebook want a cut too? Would such a thing be allowed at all? Would bots on iOS be able to do the same things as bots at Facebook.com? I can foresee bots replying “I’m sorry, I can’t do that here but if you tap this link I can help you at Facebook.com”. Everyone loves unhelpful robots who redirect you to their website. ↩
As much as I enjoy Telegram and Slack, the so-called “slash commands” are not really easy to remember, and are certainly are more difficult to discover than similar actions in apps with buttons and menus. Even for geeks like me who love Markdown it is most common for us to have various tools to automatically or semi-automatically generate the necessary–even if minimal–markup. ↩
One less obvious benefit to giving Siri an iMessage thread might be that it’d show concretely why end-to-end encryption is so vital in the service. Presumably, Siri would be able to spend money for me, and nobody likes the idea of insecure banking. This is something I think any judge or jury would instantly grasp. ↩