The AI art of the possible: What the world’s most interesting (and useful) chatbots have in common

Wondering how to choose the best gift for the holidays? Ralph is here to help. Ralph is a Facebook Messenger bot that LEGO launched a few years ago to ease the number of incoming messages that customer care agents handle during the holiday season.

Ralph starts out with simple questions: How old is the person you’re buying the gift for? What’s your budget for the gift? What region are you ordering from? (By making the bot available in multiple countries and languages, LEGO expands its reach to customers across North America and Europe.)

Based on how customers answer these initial questions, Ralph gets a bit more specific, determining the theme of the gift: travel, adventure, fantasy/sci fi, city planning, and much more. From there, Ralph recommends the perfect gift, and even adds it to shoppers’ carts for ease of purchase.

Ralph isn’t just whimsical and fun for shoppers; the bot provides serious ROI to the brand and real help to customers. The bot uses a linear decision tree to narrow down results and give customers a clearer idea of what they want. With each answer a customer gives, the bot filters out less relevant products, so that at the end of the decision tree the customer is left with a smaller, more easily shoppable list.

According to Sumo, LEGO has seen a 6X return on ad spend in certain areas, and has reduced their cost per conversion by 31% (vs. other conversion-based ads). We typically think of bots as assisting with customer service — that is, stepping in when things go wrong to ease the burden on support agents. But in this case, LEGO has creatively applied the technology to help with sales as well, and has seen excellent results.

Duolingo, the internet’s best way to learn a new language for free, is using AI in a creative way to solve a problem that’s existed ever since people began learning different languages. The problem is that the best way to learn a new language is to practice with a native speaker, but many of us (especially in the US) don’t have regular access to one. So, Duolingo created a group of chatbots to talk to users in one of three languages: Renèe (French), Roberto (Spanish), and Ada (German). They’re planning to add even more bots to the team, to help users with more of the twenty+ language courses available on duolingo.com.

In many ways, this is a standard use of AI: tackling a problem that typically requires huge amounts of time for humans to solve, thereby increasing the efficiency of the humans it serves (in this case Duolingo’s customers). But unlike so many other chatbots out there, Duolingo’s bots aren’t there for customer service or sales; they themselves are the product

Of course, Duolingo could have stopped there and already more than qualified for inclusion in this article. The use of AI as an integral part of their product, and not just as support, is brilliant.

But Duolingo has gone even further to give their customers not just a product, but an immersive experience. Each chatbot has its own character, voice, and story: Renèe is a driver, Roberto is a chef, and Ada is a police officer. While this AI isn’t quite on the same level as talking to a real person (no bot has ever won the full Loebner Prize), it adds richness and depth to a process that might otherwise be dreadfully boring.

If you’re talking about great chatbots in the last few years, there’s no choice but to include National Geographic in the conversation. NatGeo’s Facebook Messenger bot, Pop Geo, helps users find what they’re looking for and generates interest in promotions and other events.

First, the bot asks users trivia questions to get them engaged in the conversation:

Then, Pop Geo moves into promoting books, National Geographic television shows, and more. According to ChatBotsLife, results were impressive, with 65% of users coming back to the bot to answer daily trivia questions. 29% clicked through to the additional content the bot suggested, and 43% opened promotional messages — both massive improvements over these rates for email campaigns.

Still, the most interesting aspect of NatGeo’s Facebook Messenger bot is that its identity isn’t always the same. That’s because of their popular Genius television series, which takes season-long, biopic-style deep dives into the lives of important people throughout history. So far, the show has featured Albert Einstein, Pablo Picasso, and Aretha Franklin.

When new seasons come out, National Geographic builds anticipation among viewers by changing the chatbot’s identity to match the subject of that season. They use details from the person’s life and personality to craft an immersive chat experience that, of course, points the user toward viewing the show. This year, people chatting with National Geographic on Facebook Messenger were actually chatting with the bot version of Aretha Franklin!

This sort of versatility on only one platform shows that National Geographic is characteristically pushing the boundaries of chatbot technology, and we the audience couldn’t be happier about it.

Many bots get discontinued, aren’t implemented by a big brand, or are just downright too creepy to produce. But that doesn’t mean they’re not creative and interesting. Here are a couple of my favorites in that category.

Not every chatbot comes from a major brand looking to support its business objectives. Some simply come from people who want to help solve issues, like insomnia. The Insomnobot 3000 (love the name, by the way) is a chatbot designed to keep users company when they can’t fall asleep. The bot’s designers cleverly made it most “active” between 11pm and 5am — a lot like a real person who has insomnia.

Okay, let’s get (a little) creepy. Microsoft recently patented a chatbot that simulates talking to a dead person. You read that right. Per CNN, here’s how the bot would work:

...the tool would cull "social data" such as images, social media posts, messages, voice data and written letters from the chosen individual. That data would be used to train a chatbot to "converse and interact in the personality of the specific person." It could also rely on outside data sources, in case the user asked a question of the bot that couldn't be answered based on the person's social data.

The only catch? It’s just too weird. Microsoft confirmed that they have no plans to actually develop this idea into a product, mostly because it’s so disturbing. Still, though, it’s amazing to see what might be able to happen when AI is combined with powerful analytics from social media and elsewhere.

So, what do these three amazing bots have in common? The answer is creative implementation. None of them conforms to the prototypical chatbot use case, and they each come with a unique purpose and execution. These bots show us just what’s possible when we combine the incredible technology of artificial intelligence with good old fashioned human ingenuity.

A chatbot can be incredibly effective at completing simple to moderate tasks, saving brands time and money while bringing them deeper insights than humans could on their own. Chatbots don’t replace humans — no bot mentioned in this article tries to do that — but they can augment human capabilities in important ways. The only limit is your creativity.

What does this mean for your brand? I suppose you’ll have to get creative to find out. If you liked this article and want to learn more about implementing a creative chatbot for your brand (or even improving upon your existing bot), check out our ebook, The Bot Balancing Act.