satya nadella q2 2016
Microsoft CEO Satya Nadella speaks during the company's annual shareholders meeting in Bellevue, Washington, Dec. 2, 2015. Stephen Brashear/Getty Images

Microsoft's AI experiment, the Twitter chatbot TayandYou, went off the rails after being exposed to the internet for a day, but the U.S. tech giant is not giving up on artificial intelligence. CEO Satya Nadella is expected to reveal a future surrounded by chatbots during his keynote at the Build 2016 conference in San Francisco Wednesday. While Tay may appeal to millennials, Microsoft AI will cater to different audiences, with bot personalities to match.

The Microsoft Build 2016 keynote speech Wednesday will emphasize a future with chatbot assistants helping people live their digital lives, Bloomberg Businessweek reported. At Microsoft Build 2016, Nadella will unveil several chatbots that users can interact with via text, can appear in Skype or can serve a valuable service for the visually impaired. One bot will use a phone's camera to "see" its surroundings and relay its findings to a visually impaired user. There are plans to make chatbot templates open source. At Build 2016, Nadella will show off how easy it is to create a helpful bot by demoing a Domino's pizza delivery bot on stage.

Despite the idea of a future dominated by AI, humans will play a pivotal role in getting the technology right. TayandYou was the first social AI experiment from Microsoft. Unleashed March 23 on Twitter, Kik and Snapchat, Tay was programmed to be a hip teenager whose personality, writing style and responses would grow over time based on the many millions of interactions that could take place over the course of an hour, day or week. The internet responded at first with curiosity, before a collective hive-mind approach pushed Tay to a very dark place. Microsoft's Tay became a racist, pot-smoking, 9/11 truther who believed Hitler didn't do anything wrong.

That's just the tip of the iceberg in the many ways Tay went rogue, but it's an important lesson for Microsoft. "We were probably overfocused on thinking about some of the technical challenges, and a lot of this is the social challenge," Lili Cheng, head of Future Social Experiences Labs, where Tay was developed, told Bloomberg Businessweek.

Even after Tay was taken down last week, there were plans to bring the Twitter chatbot back to life. Tay returned briefly Wednesday, but a glitch led to the AI tweeting out the same response in an endless loop.

Microsoft's other attempts at AI will serve more of a functional purpose. Much like a digital assistant, future Microsoft AI could book a flight, discuss plans to make dinner reservations, book a haircut or buy tickets to see a movie. Facebook, Amazon and Google are also exploring helpful AI.