Microsoft announced its Bot Builder platform two weeks ago, Telegram announces the new version of theirs today, Facebook is set to launch the bot store on F8 this week and according to some this it’s Zuckerberg’s biggest launch since the Facebook Wall, god dammit even us at Wieden + Kennedy (an advertising agency) we are building bots for some of our customers, and I’m not going to be surprised when Snapchat reveals it’s bot market to help you achieve the perfect face swap.
Let me try to quickly explain what’s behind them, why are taking over, why they aren’t a fad and why they matter:
How? Artificial Intelligence is actually intelligent now.
If you just woke up from a year long hibernation, allow me to quickly recap what you’ve missed, computers are smart now, very smart, and they are flexing their muscles in every single field, from the coveted position of AlphaGO for being the first computer to “beat” the best of our kind at a match of “Go” (believed by some to be one of the last frontiers of human vs machine dominance) to fantastic computer vision / image recognition advancements through visual-recognition models: like the one from Andrej Karpathy and most recently Facebook unveiled it’s tool to aid visually impaired people.
Most of these achievements are being powered by the latest development in machine learning, a device called Neural Networks (NN), in all of these cases NN techniques are applied in combination with other methodologies such as Monte Carlo Trees of search, or by generating more complex learning structures such as Recurrent, Convolutional or even Multimodal Recurrent Neural Networks.
Big Data
I’m not talking about the 2012 marketing fad, I’m talking about the large datasets being released by large corporations or crowdsourced by “the people” and accessible to anyone, today you can find any information about a place in Google Places API or in Foursquare API, you can compare an image against the COCO Caption Dataset with more than 100k images to figure out what exactly is inside that picture of yours, you can compare health and DNA information with the data from 23andMe to identify people with extraordinary ability, you can ask for help from Wolfram Alpha to solve mathematical problems or ask Wikipedia to define almost anything, etc etc etc. But more importantly, with these large chunks of data, you can train a system in order to successfully predict a common outcome or to learn how to respond to substancial number of scenarios, we live in the era of information abundance, massive datasets ready to be processed, compared, analyzed and served to your users in much less than the blink of an eye.
Why chat bots?
Text is the first frontier for AI to conquer in the consumer market, and it’s probably the most important one, the question is not IF chat bots will succeed but when* and who* will own the first platform that can successfully “trick” us to interact with a machine as we would with other humans, by simply talking.
Big tech companies, know that whoever get’s to build what Apple Build in it’s App Store will have tremendous leverage to shape what the near future of technology looks like, more importantly, they’ll own the relationship between developers and hungry costumers waiting for their life to be simplified by an assistant.
Cool products such as Luka are popping in the App Store as standalone apps, but I foresee a gloomy future for these efforts. Why would you build an interface if you could seamlessly add a product into an existent platform? In other words why would you build an app when you can talk to a service as if it was your friend on Facebook or WhatsApp? Why would a human change its behavior to get things done by a machine if it could just talk to it in natural language. The era of computer illiteracy is almost over.
Why is this relevant?
Remember that time when your aunt asked for your help to buy something on Amazon? Or when a friend asked your opinion about a digital product or how to use it? Those times are soon to be gone, instead of asking for your help, next time your friend will ask a bot who’ll effortlessly recommend a product based on which stuff fits him or her better, what’s healthier, or what’s cheaper. Natural language processing represents broadening the market penetration for digital businesses, in other words: it’s about to get even easier to consume products via a digital gateway. For companies like Facebook and Telegram, it represents a new revenue stream finally detached from advertising, it represents a shift to a more Apple-esque business model, owning the platform where you consume digital goods.
Bots represent the present but they allow us to see a much more interesting picture of the future, one where humans and machines interact with each other seamlessly as equals, today you have them embedded in your messaging platforms, but tomorrow the same bots you used to text will live in your kitchen and your office.
My name is Mauricio Ruiz and I’m a creative technologist based in New York City, if you enjoyed reading this short article, please hit the recommend button!
