Don’t confuse (smartly) scripted messaging with AI-powered bots.

Since the opening of the Facebook Messenger platform to “bot” developers at F8 in April, there’s been a rush of bot designers swiftly crafting all sorts of Q&A/IVR-type flows delivered via the Messenger chat interface. But we shouldn’t confuse (even smartly) scripted messaging with AI-powered bots.

Actually there’s been a mainstream AI-powered bot out there on the market for ages, it’s called… Google. It’s a proper bot in that sense that it’s aimed at automatically processing the natural language of your search queries (and it does a pretty good job, incl. for voice triggers) to display the most accurate answer it can provide (even tailored to your own personality), ideally without the extra need to click on the results.

There are no human researchers on the other side of the screen mining the worldwide web to find the answer. Google is a fully fledged autonomous bot, everyone’s #1 personal assistant.

And, as you might have noticed, it’s been improving a lot since the early days. For a growing selection of specific use cases, you can get results displayed as a nicely structured widget: for Wikipedia-type queries (Who’s the President of the United States of America?), weather (What’s the weather now in London?), sport results (Who won the match between England and Slovakia?), calculations (How much is 100 divided by 20?)

voice search — sport results
voice search — calculation

In most of those cases, Google (voice) search beats any form of messaging interface, which would require multiple steps to come up with a similar answer. Google is probably the most efficient consumer facing bot on the market today (my experience with Siri has never proven as successful). We should also pay attention to what the guys at Viv (the Siri founding team) are building, pretty mind blowing.

We’re also using an infinite number of invisible bots on a daily basis, i.e. the increasingly powerful algorithms powering all sorts of apps. Each time A.I. is at play to deliver a structured output based on your deliberate input(s) and/or your behavior, you’re interacting with a bot, even if this interaction doesn’t have a conversational format. Questions & answers pretending to mimic a human conversation aren’t always the best way to deliver the expected result in an efficient manner.

So yes I do firmly believe in the advent of bots becoming smarter and smarter, learning from us, their context and the internet at large to deliver relevant results in an automated way, but I think we shouldn’t fall into the trap of thinking that the best form factor to engage with those bots will always be chat / messaging as we know it.

Facebook Messenger as a platform could become — thanks to its unparalleled 1B+ reach — an extremely powerful distribution channel for the new breed of bots but, in my view, its ultimate success would require to extend the interaction possibilities beyond the sms-inspired messaging format.

Bots usually work best when you don’t notice what’s going on behind the scenes, hence the seamless success of Google, that few people are actually calling a “bot” when it’s probably the best proof of concept for the automation of knowledge processing at scale.

By the way, I’d love to invoke a Google bot from a Facebook Messenger conversation with my friends. Poke to Mark, David & co ;-)

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade