Despite the hype, chatbots are in a precarious position. Many consumers have interacted with one in some guise – but they’re not actually something that the consumer world at large needs.
Often, they’re a marketing gimmick to drive cheap engagement. That’s not to say that there are no success stories: Intercom’s chatbots, for example, are gamechangers. In other cases, however, they exist in a dusty corner of cyberspace, slowly going obsolete like USB flash drives and landline telephones.
That said, many analysts are optimistic about the future of chatbots, putting the tech in a tentative infancy. According to Juniper Research, chatbots will be responsible for cost savings of over $8 billion annually by 2022, up from $20 million in 2016 – so will 2018 finally be the year of the chatbot?
The good, the bad, and the ugly
The true potential of any technology is in its actual uses – and chatbots have seen wide proliferation, snagging a multitude of headlines along the way.
Microsoft’s AI-powered bot, Tay, famously made headlines when a barrage of users engineered the bot so it would tweet out racists slurs. The bot responded to tweets and chats on Groupme and Kik but was quickly shut down for its nonchalant offensive statements.
Tay was able to tell jokes, offer comments, and complete simple conversational tasks. But this being an open forum on the internet, users quickly taught Tay racist language, including Holocaust denial and siding with Trump’s plans to ‘build a wall.’
Tay displays the best and worst of chatbots (and internet users): it was a showcase for conversation but its machine learning algorithm was derailed by a lack of context.
As high-profile a failure as Tay was, other chatbots are succeeding with pace. In Hong Kong, for example, they are serving tech-savvy customers and cutting costs for banks.
HSBC bank and Hang Seng Bank are releasing three chatbots between them: Amy, and HARO and DORI respectively. Over the last two years, the Hong Kong Monetary Authority – the city’s banking regulator – has encouraged banks to develop financial technologies to cut down operating costs, and banks believe chatbots may be a big part of the solution.
The chatbots will handle basic customer queries, with a chatbot inquiry estimated to save around four minutes compared to a traditional call centre. HARO will work via the bank’s app, while DORI will be housed on Facebook Messenger.
Which leads nicely onto Facebook and its own chatbot projects. In mid-2017, Facebook’s AI researchers were tasked with teaching chatbots how to negotiate. However, the bots were left unsupervised and developed their own machine language. As the bots were intended to communicate with people, they were shut down for ineffectiveness.
Facebook’s other big chatbot project, M, saw a similar fate. The chatbot was a virtual assistant which used a combination of human contractors and AI software to answer queries and perform tasks in Facebook’s Messenger app (for example, buying a product).
Curiously, as Facebook winds down its chatbot activities, Amazon and Google are ramping theirs up in the race to create the most fully-realised voice-assisted virtual assistant.
Why voice and search are the missing link to chatbot efficacy
In 2017, 35.6 million Americans used a voice-activated assistant device at least once a month – up 129 percent compared with 2016.
The true power of voice search, however, isn’t in phones or computers – it’s in personal assistants like Echo and Alexa. Amazon dominates the market with 70 percent market share, with Google in second with a paltry 23 percent. Obviously, Google powers the Echo, while Microsoft’s Bing powers Alexa.
Like many new technologies, chatbots will live or die by their usefulness. While banking, retail, and customer care have seen wide-scale take-up of the tech, expect increasingly smart machine learning to bridge the gap, with virtual assistants ushering in a new frontier for chatbots.
Likewise, the tech powering the chatbots themselves will get smarter with natural language processing developing to better recognise opinions, moods, and sentiments in online conversations.
With improved NLP, expect chatbots to spread to far-reaching corners of the internet. In a nutshell: 2018 may well be the year where chatbots finally enter the mainstream.
Interested in using AI to power your customer services?
If you’re interested in finetuning your customer relations services, Statwolf can help. With more and more companies looking to chatbots, NLP, and sentiment analysis, you’ll need an expert team on your side to keep up with your competitors.
Our advanced data science consultancy can team up with you to unlock the patterns in your data so get in touch today to unleash your business’ potential. We offer a free Machine Learning Consultation too so feel free to avail.
Want to make sense of your data? Download our comprehensive guide: The Predictive Maintenance Cookbook.
