Last week, Microsoft launched @TayandYou, a “chat bot” designed to engage people on Twitter. The company’s Technology and Research teams claimed that chatting with Tay would make it smarter, thanks to the wonders of machine learning.
But the combination of A.I. and Twitter proved a volatile mix: Mere hours after Tay launched, hundreds of merry pranksters figured out how to make the bot spew racist, misogynistic drivel. As Tay “learned” from its new friends, it quickly devolved from bubbly teenager to the online equivalent of an SS colonel with Tourette syndrome, at which point Microsoft mercifully shut the project down.
Despite that high-profile failure, Microsoft reportedly intends to release still more chatbots in coming months, according to Bloomberg. Unlike Tay the Twitter novelty, these bits of semi-autonomous software will help users perform a variety of mundane tasks, whether booking flights or sending automated messages to colleagues.
Bloomberg predicts that Microsoft will show off some of these bots during this week’s BUILD conference in San Francisco. The company also plans to issue tools that allow anyone to custom-build a bot for their own needs. If you’re a small e-commerce firm, for example, you’ll be able to create a bot that facilitates a customer’s ability to order.
In essence, bots represent Microsoft’s opportunity to advance the tech industry beyond its current obsession with apps, and (hopefully) eliminate at least some of the complexity that clogs up most folks’ daily lives. But that all depends on avoiding another Tay-style debacle, given how consumers and businesses are quick to dump software that goes haywire.
Other tech firms have already begun to exploit the potential of bots. Slack, arguably the most popular enterprise-chat platform on the market, features tons of them, most designed to make tasks more efficient.
For tech pros, the rise of bots could present a whole new sub-industry. While it’s an open question whether bots will someday become a ubiquitous as apps, the increasing sophistication of machine-learning and artificial-intelligence platforms all but guarantees that more people will rely on “digital helpers” in years to come. If you’re interested in building bots, though, just make sure you don’t produce another Tay.