How close are chatbots to pass Turing test?
Kumar Shridhar

Do all chatbots need to have AI? I think the perspective about what a bot is supposed to be has been a bit distorted by what grabs better media headlines (Artificial Intelligence).

While people are struggling to make a bot that passes the touring test, I’m more interested in bots for the following reasons.

  1. Frictionless Discovery. More than half of US smartphone users download 0 apps per month [1]. It takes a lot of convincing for someone to download your app, register a new account with an email, and learn the interface. Instead, just chat to a bot instantly in the application you and your friends are already using (Kik, Facebook, Telegram, Slack).
  2. Ease of development. With the Kik or Facebook API’s, it is trivial to create a chatbot in an ecosystem exposed to millions of users; all with very little resources. I’m not a developer, but I was able to build a bot with Python that’s reached nearly 2 million users so far. [2]

AI will be great eventually, but until then we could solve a lot of problems from simple, easy to develop, low friction interfaces within our favourite messengers.

@uber: pick me up

@weather: whats the weather tomorrow?


One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.