Artificial Intelligence (AI) is going to change the world. When this will occur depends on whom you ask. What we know for sure is that it will happen and profoundly affect society. We also know that it is going to be a gradual process, as is often the case with technology and innovation. The process has already started though and proof of its spreading reach is becoming evident.
We, humans, are imaginative beings. We saw the moon and its beautiful glow at night and thought, wouldn’t it be great to land on that rock, quietly observing our planet from up there? Sure enough, we pulled it off.
Give it sufficient time and the AI possibilities we see on the horizon today will be reached as well.
AI is already here
Many people are excited about this prospective, whereas others are worried about job displacement or downright frightened by the idea of something much smarter than humankind replacing us. (I expect Hollywood to continue to wildly speculate upon such scenarios for years to come.)
AI will likely have to be regulated to help keep the risk of these doomsday situations to a minimum. Personally, I’m more in the enthusiastic camp, though. I don’t see Artificial Intelligence as an outright replacement for humans – at least not in the span of our lifetime.
AI’s big impact
What excites me is its potential to augment what humans can already do, in both small and big ways. Let’s start with a big example: doctors reaching faster and more accurate diagnoses when assisted by cognitive computing. This is particularly true for rare conditions and diseases whose early diagnosis may be crucial for survival. In such an instance, AI would be a game changer.
And if you think that would have a huge impact, consider for a moment self-driving cars which are significantly safer – even at this early prototypical stage – than human drivers.
Globally, over 3,000 people die each day from car crashes, according to the WHO. An additional 50,000 – 135,000 people are injured or disabled every day by auto accidents. These are staggering statistics that affect many millions of people each and every year.
It would be hard not to look forward to a future where AI can save and improve lives to such a dramatic extent.
AI’s small, but pervasive impact
However, it’s not just big things that can be aided by this technology. AI is quickly seeping through the seals of our lives in small but meaningful ways. Spotify and Pandora both use it to enable us to discover new music that we may very well end up loving; Netflix, to predict and suggest what we might enjoy watching.
It is leveraged by retailers to better understand our preferences and predict what we may be interested in purchasing based on our online habits and past purchases (occasionally in a tad creepy way).
Smart IoT devices are increasingly capable of learning our behavior and adapting to it. For example, the Nest thermostat (now owned by Google). It helps homeowners conserve energy and save money while optimizing comfort levels according to their daily habits and preferences.
Perhaps one of the most pervasive and fastest growing applications of Artificial Intelligence to our daily lives is the emergence of virtual assistants à la Apple Siri and Amazon Alexa. These are both examples of chatbots.
A chatbot is a software program that leverages AI to have a conversation with a user, usually via text or audio (like Siri and Alexa do).
The chatbot will typically offer some form of useful assistance to the user (or at least entertain them). There are chatbots for all sorts of purposes, from booking trips to helping people who are struggling with depression (and everything in between). We are still in the early days of chatbots, but I would argue they are already the most common form of applied AI out there today.
Gartner predicted that by 2020, 85% of interactions between users and enterprises will be through chatbots and related technologies. 80% of companies will want a chatbot by then. So much so, that there will most likely be more chatbots available than mobile apps!
Chatbots are a win-win
Why so much interest? The business case is clear. The idea is not so much to replace your customer support team outright, as it is to provide a better experience with immediate, 24/7 answers to your customers in a wide range of common situations. (With escalation to a human team for more complex scenarios.) A chatbot can provide customer support (and even pre-sales support) in a pop-up chat window on your site, through platforms like Facebook Messenger, or even via SMS.
Consider a hotel chatbot. It could easily handle queries about hours of operation for various services and facilities, such as a restaurant, room service, pool, gym, check in and check out times, breakfast policy, airport shuttle schedules, and much more.
While the chatbot does its magic, the humans working in reception and customer service can focus on more complex queries. For example, handling a guest who found dozens of live crickets in their room. (This actually happened to me once, but that’s a story for another time.)
How a chatbot works, in principle
Conceptually, a chatbot is rather simple. It’s a software program that receives a message from the user, interprets what the user requested, and then provides a meaningful response in output.
The tricky part is having the software understand what the user’s intent is and then determining what germane information should be given to the user (or what follow up questions should be asked to clarify their input.) This back and forth interaction should feel natural and somewhat human-like, though chatbots should never pretend to be humans.
That’s where AI enters the picture. If you can leverage an engine for the cognitive heavy lifting required, designing a chatbot becomes significantly easier. And that’s exactly what Watson provides you with via the IBM Watson Conversation service running on the IBM Cloud (a free platform with premium, pay-as-you-go options).
Anyone can create chatbots
The good news is that you don’t need to be a PhD computer scientist to create chatbots. In fact, you don’t even need to be a programmer.
Ask Allan Shulman of Dollar Tea Club. A business owner, not a programmer, he quickly learned the skills required to build a chatbot via Watson Conversation. Guess what? It’s already saving him time and money thanks to a reduced number of support requests. I firmly believe that most business owners can do the same for their businesses.
When you create a chatbot with Watson Conversation, you define intents, entities, and visually arrange a series of nodes that form the dialog. Intents capture the user goal or purpose. For instance, you feed some examples of what a #greeting intent might look like to Watson, and it will learn to recognize when the user is greeting the chatbot (even when they phrase the greeting differently from the examples you’ve trained Watson on).
Entities will capture specific details of a conversation. Think a date and time of a reservation, the user’s name, or the specific facility the user is asking about.
Finally, the dialog will define nodes to handle specific scenarios. For example, one node to handle greetings and respond appropriately; one to answer questions about, say, restaurant hours, and so on. The dialog will shape the interaction with the end user.
Since this is done visually, and Watson handles the Natural Language Processing (NLP) part for us, you don’t need to know how to code to create chatbots with Watson Conversation. Of course, existing programming skills don’t hurt when organizing the dialog. It can also allow you to take the chatbot further by integrating other services and external data sources.
Deploying the chatbot
Once you’ve created a chatbot that understands the user reasonably well within the scope of what a chatbot is supposed to do, you might think that programming becomes necessary.
After all, you’ll need to create some form of a pop-up window on your site, and figure out a way for the input text to be sent to your Watson Conversation service. In turn, you’ll need to receive its response to present it to the user. This approach (i.e., using a proxy app) is valid, but it’s too technical for non-programmers.
Thankfully, we’ve developed a WordPress plugin that allows you to deploy your Watson Conversation chatbot in a few clicks. No coding or programming skills required. You install the plugin, enter your Watson Conversation credentials, et voilà you have a chatbot running on your WordPress site. Seventy-five million sites use WordPress. It’s the most popular Content Management System (CMS) in the world. We felt that it was the right platform to help business owners like Allan, create and deploy useful chatbots.
You can also see the plugin in action on Cognitive Class, a free data science and AI learning initiative by IBM. If you click on the chat icon, you’ll see a pop-up in the bottom right corner and you’ll be able to chat directly with the chatbot itself. While I have you on Cognitive Class, consider enrolling in my free course, How to build a chatbot. Within it, I provide a gentle introduction to chatbot building that specifically assumes no technical background.
I also cover deployment on WordPress, so you’ll be able to take the chatbot you created and deploy it on your site. Give it a shot. I look forward to seeing what you build.