Some thoughts on bots & AI in business

Posted by kdow on Nov 21, 2017 11:28:02 AM

Photo by Andy Kelly on  Unsplash

When I studied computer science we had some courses that covered the core concepts behind AI. This was before “bots” did anything beyond the basic functionality of a magic 8 ball in a public IRC chat room.

These days they’re all the rage. Particularly since the accelerated growth of Slack, which allows companies/teams to implement bots-as-a-service. Most bots do simple lookups right now, replacing search bars in browsers. One simple example that was launched by a colleague yesterday allows sales reps to look up case studies & competitive information on Slack. This bot follows a simple tree-sorting system that guides the user through a predetermined flow to end up at a URL or answer in String form.

Most bots seem to follow a similar pattern. However, AI introduces the idea of machine learning & complex data flows to allow a more fluid flow of information, dictated by some logic stemming from natural language processing rather than predetermined flows that feel obvious to the user.

A normal bot, like the one my colleague launched yesterday, often has some UI akin to buttons. Something powered with AI can be perceived as intelligent, with the UX being more akin to an actual chat window. The level of intellect the bot can ape is dictated by the algorithms defining it’s ability to chat as well as the data it has to hand (both in the form of tree-sorting algorithms with responses that lead to the answer as well as the context provided by you as the user).

If you’ve not played Akinator, it’s a fun example of tree-based logical decision making at a large scale. Akinator is almost uncanny in it’s ability to figure out a celebrity that you’re thinking of. But it’s a rather simple system, built at a huge scale with lots of good data.

In the business world, the Akinator that helps a sales rep, marketer or anyone really would be GrowthBot.

But this isn’t the future of bot-driven business lookup or decision-making assistance. No, the future would be something that has an unimaginable amount of data to hand, with thousands of users interacting daily — with the bot then using said interactions/answers, etc. as a way to crowdsource a better way to communicate and deliver results. Machine learning to provide the right context & answers, with the AI engine providing the ability for our bot to converse in a more human way.

What I’m getting at is the idea that, if a bot is to truly revolutionise the way people work and interact with digital information it would need to pass the Turing test. The Turing test, for those unaware, posits that a sufficiently powerful bot/AI can only ape human cognitive ability when compared to other humans. So the subject (i.e. the bot) is pitted alongside a human in a chat window to converse with a third party human who must determine which chat window is one with a human at the other end of a keyboard, or a piece of software. Should the third party human decide the bot is a conscious being, then the software has passed the Turing test.

Keep in mind Alan Turing (from whom the test is named after) was a gay man in a time that gay men needed to hide in plain sight. In fact, despite his contribution to British society & the creation of Computer Science, he was chemically castrated due to his illegal practice of homosexuality. The test for a bot is almost the same as it was for a gay man in the 30s, 40s & 50s. Hiding in plain sight. If a bot can ape human ability through a chat window, delivering insights to the end user as if a human was doing it, then we can truly consider an AI/ML revolution taking place.

And don’t get me wrong, I still think Blockchain will be more revolutionary (heck, it might be the backbone of smart bots) in the long term, but right now, we’re in the midst of a digital access revolution thanks to more powerful AI computation and Machine Learning having access to far, far more data than ever before.

Bots aren’t only a way for sales reps to find case studies, though. Imagine poorer, developing regions accessing bots via SMS to find the best way to treat common illness or find the nearest education centre for their kids. Imagine a bot being used to solve complex customer queries of a technical nature, or fulfilling orders on an eCommerce site during a busy period (xmas, for example).

With good access to data, proper AI mechanics and a proper use-case, bots could be a huge digital revolution. Right now we’re in the nascent stages of this revolution.