“Let everyone in the office know I’m running late.”
This simple statement could change everything about the modern workplace.
It’s not possible for an artificial assistant to execute this command today, for several reasons — issues with calendar programs, subpar artificial intelligence, or even the fact that there are so many different definitions of everyone. A chatbot today — like Siri or Google Now, or even Amazon Alexa — could almost handle this request, but they are not quite advanced enough. For starters, how does Siri know who is in the office? You’d have to have a check-in system in place and the check-in system would have to communicate with Alexa.
There are also some problems with contextual understanding. Running late for what? The weekly status meeting or a coffee date? Is it the next meeting on my schedule or is it the next meeting with everyone at the office? An A.I. can’t quite understand the language I’m using because we like to talk to bots in the same way we talk to each other. We’re nonspecific. We say one thing and mean another. If I’m running late to the office, it’s obvious to most of us that I’m talking about that meeting, but the words don’t really translate to a chatbot with a limited understanding of what it means to be human.
Lately, I’ve been thinking a lot about chatbots, because I’ve been testing them like crazy. I booked a flight with Lola, I found a restaurant that serves cheese curds using Ozlo, I ordered a Bill Bryson book using the Mezi app. Unfortunately, of those three apps, only Ozlo uses 100 percent artificial intelligence, and it’s by far the most limited in what it understands. For the dozens and dozens of chatbots I’ve tested here at VentureBeat in my role as an editor for our bots coverage, I’ve come to realize that many of the bots are helpful and useful, but they have a long way to go before I’m ready to hand over my schedule, have them parse my email (or reply to my boss when he sends me an urgent message), or even order me a burger for lunch.
Why is that? From what I can tell, chatbots are in the same stage as voice recognition software when it first became popular quite a few years ago. As more people talked to their phones and computers, the recognition software recorded more dialects, understood more hidden meanings, and built up a vast dictionary of natural language understanding. We can now talk to Alexa using everyday speech, thanks to the pioneering work of software engineers from the past, who built up an engine slowly over time and eventually improved the technology to the point that we are hardly aware of how complex it is.
This is what makes chatbot innovation such an exciting field. It has so much promise and so much potential. Someday, a chatbot will handle most of the requests we receive from strangers (this is already partially true with the MessinaBot). That means we won’t have to field so many random questions by email, which is worth its weight in gold if you share my views on email overload. A chatbot will slowly learn our preferences — that we like to meet later in the day or that we don’t like Microsoft so we’ll never respond to an email from someone trying to convince us to upgrade to Windows 10.
A chatbot will know our likes and dislikes, where we like to work when we’re not in the office, which car we like to drive, and even which restaurants we frequent on Sunday afternoons. For this to happen, chatbots will need to connect to more than a Gmail calendar. We’ll need to live in a highly connected society. Think about how a chatbot would even figure out which car we like — it would have to know we physically visited an Audi dealership three times in the last month and that we own an Audi right now.
Yet, we’ll get there. Just like speech technology, chatbots will only learn more and more about us and connect to more and more systems, other apps, and even other chatbots.
I can’t wait for this chatbot revolution.
For now, hang on — I need to text everyone in the office manually to let them know I’m running late.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.