It’s very likely that you’ve come across phrases such as automation, machine learning, artificial intelligence and ‘chat bots’ lately. There are quite a few reasons for this, the main one being that in recent years - 2016 particularly - there have been some exceptional leaps forward in thinking and capabilities, as well as some real world advantages to making use of them.
To start with, let's look at a bit of the background to this area. Going right back to the very earliest home computers, the idea of ‘conversing’ with your computer has always been a bit novel but a ‘fun’ example of an almost-but-not-quite realised 'science fiction’ ideal. Simple programs where you could respond to simple prompts, and ask basic questions were actually common place; but very ‘dumb’ in that the responses were limited to what had been pre-programmed.
Over the years, with the growth of the world wide web, it has become possible to write the same kind of programs into a webpage. This has given rise to fun ‘chat bots’ that you can ‘talk’ with via a web page. They were primarily for fun and general interest, but from 2017 things have changed drastically.
Evolving largely from research and development, spearheaded by advances in search engines especially, it's become possible for applications and ‘code’ to parse, and in a sense understand, input text, regardless of whether that text comes from a web page, or a person typing into a text box. Even more excitingly, through machines learning algorithms, neural networks and natural language processing, very intelligent and relevant responses can be looked up or produced very quickly, to the point where it is even possible to be fooled into feeling like you are talking with another person.
Besides a bit of harmless fun though, what use is this? Well, in the same way that we’ve been engaging with search engines to find answers to questions and discover content or products that is relevant to what we are looking for, the same process can now be refined and made more specific by employing these ‘AI’ techniques in all kinds of novel ways across different areas.
Besides ‘Cleverbot’, an online chat bot that grows increasingly capable constantly, by learning from the many users and conversations that it is exposed to, a good example of how far things have come is IBM's ‘Watson’ AI. Watson, in ways an evolution of the Deep Blue AI that beat a grandmaster at Chess, is an AI backed by data from various encyclopaedias, dictionaries, thesauri and websites, particularly online news sources. As well as having won on the TV show Jeopardy!, Watson has been successfully used in healthcare as a decision support tool, with hopes of eventually being part of a clinical diagnosis process. Watson is also being integrated with ‘chatterbot’ - which is a program used to provide conversation in some children's toys. And is also being developed to be able to provide weather forecasting.
Most exciting and relevant of all though is the fact that Watson, and other similar systems, are being used to assist shoppers in ecommerce scenarios. The term that has been coined for this is ‘conversational commerce’ - where a chat interface on your commerce site, and ultimately perhaps more general purpose messaging apps, can allow 1-to-1 conversations designed to guide the user through purchase decision making processes, and provide extra information and context in order to ultimately increase conversion rate and buyer confidence. Just a few areas that a ‘chat’ scenario can help with include:
- Cross-selling and upselling
- Offering discounts and enticements
- Receiving feedback
- Assisting with decision making
- Answering queries and offering reassurance about stock levels and delivery dates
All logged and recorded, constantly refined and improved, and without complicated interfaces, or deploying actual human beings! Even better, this can be achieved through familiar chat interfaces - even popular ones like Facebook messenger.
This use of automated ‘AI’ solutions can be far more cost effective than employing and training a large team of people, previously the only means to achieve this kind of retail experience due to technical limitations.
Here's a few examples in depth:
Another great example of just how far things can be taken is the popular chinese app WeChat, where anything from ordering a taxi, to having goods delivered, or booking a restaurant or appointment are all assisted by AI ‘chatbot’-like features. WeChat is often described as every service you could possibly require online, rolled into one intelligent app that can help you order and pay for almost any good or service, instantly.
Users of Slack are probably aware of ‘slackbot’ which is a much simpler but still very powerful and sophisticated ‘agent’ that exists in Slack’s organisation-focused instant messaging application. Slackbot can remind you of meetings and other events, alert you to certain words said in a channel, and can be tailored to provide specific helpful responses such as wifi passwords or phone numbers based on specific commands entered by users. This helps to deliver short form information that would otherwise have to be searched, with immediacy.
Wrapped up in an elegantly styled, ‘gadget’ package, Amazon’s Echo assistant takes assistants such as Siri , Google Now, and Microsoft's Cortana away from the mobile device, and puts them in your home like a member of the family. From performing actions as varied as changing the music, turning on your lights, or ordering a pizza - not to mention merchandise, apps, and media from Amazon’s marketplace - Amazon Echo has some very powerful capabilities; capabilities which being an AI, can be constantly refined and upgraded behind the scenes.
So many other examples exist, from AI’s or ‘bots’ composing music and poetry to an example of an AI writing a film script (it’s surreal for sure - but still watchable! Here's the link if you've got a spare 10 minutes!)
What happens when it goes wrong?
As with any cutting edge technology - and this really is one such area - there are always shortfalls and pitfalls along the way. One famous example is the Microsoft ‘bot’ named ‘Tay', which due to the way it employed machine learning, was seeded with a very skewed database of negative and colourful responses that it had been fed by people doing what people do best. The results, before Tay was taken offline, were some rather alarming comments that it posted publicly. Tay has now been superceded by ‘Zo’ - find out more here.
Whether you're looking to improve the interface and interact better with your customers, learn about your audience, or just have fun trying to reproduce a human experience through automated, self learning ‘bots’, it is certainly worth keeping an eye on this exciting area of technological development through 2017. Last year saw this approach overcome many technical hurdles and boundaries. 2017 might just be the year where this becomes truly refined, and commonplace; entering the home, the classroom, the office, and of course, shops of all kinds.
A final bit of fun
Aside from the fun of cleverbot and similar portals where you can get a feel for where this technology is at, here’s a lovely example of the wonderful things that can happen when two contemporary AIs are allowed to meet and converse. It turns out we are perhaps already not so far away from our Robot underlings.
And if that's not enough for you, here's some wider reading on this fascinating subject:
- 11 Examples of Conversational Commerce and Chatbots in 2016
- Wikipedia page on Watson
- 2016 will be the year of conversational commerce
- TechCrunch articles on conversational commerce
- The Guardian articles on Chatbots
- When bots go bad: common UX mistakes in Chatbot design