LawDroid Lesson: AI and Chatbots 101

VIDEO TRANSCRIPT

Hi, I’m Tom Martin and today’s LawDroid lesson is about AI and Chatbots. I’m a Yale philosophy major, lawyer, technophile, and entrepreneur. I’m the co-founder of the Vancouver chapter of Legal Hackers, and I’ve created LawDroid which is an AI legal assistant to help people solve their legal problems.

A Chatbot. In short it’s a computer program that simulates a conversation with a person and a lot of people say that Chatbots are the new apps because they’re on your mobile phone and they’re proliferating like crazy.

Well, intelligence is hard to define but in 1950 the Godfather of computers, Alan Turing, came up with a test and basically he said that if you’re able to have a five minute conversation by text and you can’t tell whether or not the other person who’s talking to you is a computer or a person, the computer’s passed the test. Actually he predicted that by the year 2000 the computer would win that test 30% of the time. Actually by 2011, a computer called Cleverbot was able to fool humans that it was in fact a human itself 60% of the time so we’ve come a long way.

Also strangely, in 1637, Rene Descartes in his “Discourse on Method,” actually came up with a very similar test. He predicted that there would be machines that look like human beings and one way you would be able to tell them apart is that they wouldn’t be able to say something that made sense. You know, close to [400] years ago you have a philosopher coming into the same test as Alan Turing.

In 1956, John McCarthy and Marvin Minsky, two computer scientists had the first conference at Dartmouth about all of this and they actually coined the phrase “Artificial Intelligence” which is the theory and development of computer systems to perform tasks that normally require human intelligence and it can be anything from visual perception to speech recognition, decision making, and translation of languages.

Some example of Chatbots as they are used today are Poncho. You let it know your location and it’ll tell you what the weather is and whether or not you should be wearing a poncho when you go out.

Another example is 1-800-FLOWERS. There’s a lot of commercial bots like this where you can order a product on your phone just by answering a few questions and chatting with the bot and it will deliver that product to you.

There is also advice bots like HealthTap where you can in natural language just describe the symptoms that are occurring to you or a loved one and it’ll serve up some answers that it thinks are relevant. It’s a great source of quick and easy medical advice.

How does AI work in Chatbots? Well typically there’s an inquiry from a user and it is parsed by a natural language processing tool like Watson for example, and it extracts the fields or parameters that it needs to match against the database of information and then serve up the answers.

There’s two types of AI. There’s rule-based AI which is essentially if-then statements. If a particular input is provided, then there’s a particular output. It lends itself to this typical decision tree-type logic.

Machine learning AI is a little different because the input may be something that doesn’t fit squarely in the box. Machine learning allows a computer to learn different responses and come to its own unique response based on an understanding of what the user is asking about. It allows questions and answers to be much more dynamic in their nature.

Chatbots are popular now because messaging apps have completely surpassed social networks in terms of popularity. On this graph you can see that the monthly active users has just jumped over social networking.

Also it’s become so cheap and easy now to create a bot, all of the major players have jumped on board. You have Google, Amazon, Microsoft, each of them have Chatbot platforms where you can design a Chatbot either free or inexpensively, and there’s a lot of platforms for Chatbots like WhatsApp, Slack, Telegram, WeChat, Skype, and of course Facebook Messenger.

The messaging app wars are fierce. There’s a huge battle for market share. This graph is a little old but Facebook Messenger and WhatsApp are at about a billion monthly active users each.

Then the other reason why now is that the ecosystem is just exploding. You have a lot of competition in this area. There’s essentially a land grab going on and if you want to get in there and stake a claim, you got to do so now.

The future. One objection I get a lot about chatbots is that it doesn’t do everything for me and so right now these are not yet the droids you’re looking for but they’re close and they’re going to get better. This graph shows where we are now in computer performance but in short order it’s going to outstrip human performance and you’ll get even better results and performance.

Lastly, just keep in mind this is an issue to keep watch of. JP Morgan recently replaced 41 years of legal work with a computer that was able to do 360,000 hours of legal work in a fraction of that time, a few seconds. Watch out for bots and AI. They’re coming down the pike but do your best to get ahead of it and get involved with Chatbots. I would suggest you check out Chatfuel which is an easy tool to get in there and get started and start playing around with chatbots.

Feel free to check us out at lawdroid.com or keep in touch with us on Twitter @LawDroid1. Thank you very much and this has been your LawDroid lesson.

Leave a Reply

Your email address will not be published. Required fields are marked *