A few months ago, chatbots made headlines for their first major failure. Microsoft's Tay, a twitter bot, learned racism within 24 hours and tweeted such absurd and offensive claims as the Holocaust "was made up." The bot was soon taken down, and yet, despite such a high profile failure, excitement about chatbots remains high today.
After all, the technology is still quite young. The potential for chatbots' growth and improvement is too great to ignore, and all the major tech companies are building them, including Google, Apple, Microsoft, Amazon, and several noteworthy startups like Viv and Slack.
So what should you know about the burgeoning world of chatbots? For starters, here are 3 interesting and futuristic facts about chatbots.
1. People are already falling in love with chatbots. Literally.
In the past year, Microsoft's Tay was a failure and Facebook's M debuted with a fizzle, but did you know that Microsoft has already released a massively successful chatbot in China? Xiaoice currently has 40 million users and chats with 15 million people daily on Weibo, China's hugely popular social network. Unlike many chatbots, Xiaoice offers no functionality to users outside of conversation. Echoing the 2013 film Her, many Chinese users have professed their love for Xiaoice, who has the personality of a teenage girl, and find a level of companionship that has never before been seen between man and machine.
While natural language processing still has a long way to go (more on that below), many users are being seduced by Xiaoice already, and the average conversation length is 23 "conversations per session" or CPS, a metric referring to a single back and forth exchange between the two conversational parties. For context, other chatbots have an average CPS between 1.5 and 2.5, showing just how advanced Xiaoice is. While many researchers believe that the future of chatbots is aiding people in accomplishing tasks, Xiaoice suggests the exact opposite: sometimes people just want to talk.
2. Chatbots could be indistinguishable from humans in 20 years.
Companies continue to sink hundreds of millions of dollars into chatbots every year in a race to answer one question: who can build a natural language processor that understands people the fastest? Modern frustrations with chatbots lie in the fact that they don't always understand the text sent to them or they provide a confusing response that doesn't really answer the original question. Making smarter chatbots is a matter of machine learning.
In short, it'll take time. Ray Kurzweil, a pioneer in natural language processing and currently working on Google's chatbot projects, believes that chatbots will be able to pass the Turing Test and be indistinguishable from humans in a blind test by 2029. "If you think you can have a meaningful conversation with a human, you'll be able to have a meaningful conversation with an AI in 2029," Kurzweil said, "but you'll be able to have interesting conversations before that."
However, this level of intelligence does come with a setback. There is a theory called the "uncanny valley" that suggests as robots become more humanoid people trust them less until they pass a second threshold and become nearly indistinguishable from humans. Research supports the existence of the uncanny valley in terms of physical representation, and some are worried that this phenomena will cross over to the context of text chat as well. As a result, many engineers include quirks and oddities with their chatbots to ensure that users know they are talking to a bot and not another person, despite the bot's intelligence.
3. Chatbots will change the market, though not in the way you might think.
"Chatbots are the new apps," Microsoft CEO Satya Nadella said in a recent presentation. Evernote founder Phil Libin believes that chatbots have as much potential to change the market as the first iPhone did. It's too early to tell just how big the chatbot market will be, but early investors and tech enterprises say it's going to be huge. It's already impacting many industries. An artificially intelligent lawyer was recently "hired" by a law firm and will give clients advice via text chat.
Digibank, a bank exclusive to mobile devices, has a staff of chatbots capable of answering thousands of different financial questions. It's clear that chatbots provide an effective avenue for businesses to slash prices by automating processes, and it's already in effect. For example, research firm Gartner believes that by 2017, only one-third of customer service interactions will require human support. Instead, chatbots like Amelie will handle most customer requests. This coexistence between worker and bot will only increase as this technology develops. Early research suggests that 12.7 million jobs will be created in the US alone to build and train robots and automated software by 2025, and that a third of the workforce will work alongside bots by 2019.