Daniel Lemire's blog

, 4 min read

We are passing the Turing test right on schedule

6 thoughts on “We are passing the Turing test right on schedule”

  1. Simon Hibbs says:

    You’re comparing apples and oranges though. In the Turing test you are knowingly attempting to interrogate the computer to determine if it is a conscious thinking being or not. That’s quite different to engaging in a conversation with a specific context and on a very limited topic.

    Furthermore the ’13 year old boy’ incident from 2014 was a very highly constrained version of the test with the rules heavily stacked in the chatbot’s favor. Modern chatbots can fulfill useful roles and interface systems, but they are a dead end in terms of achieving general purpose AI.

    1. Kenneth Stein says:

      Turing knew what he was talking about. Language understanding is grounded not intellectually, but experientially. Think about it. No, really think about that.

  2. Matt Fulkerson says:

    Besides Turing, Asimov was also a pretty sharp person. Daniel, do you believe enough researchers are putting effort into the ethics side of AI?

    Second question: Even if they are, surely not everyone is. Will ethical AI’s win out over unethical ones?

    1. Matt Fulkerson says:

      Thanks for the nice perspective. At some point I’ll read John Nash’s game theory work to see how it may apply to the question.

      Also, if people can show that ethical behavior wins out over unethical behavior (in their everyday lives), then we at least know that this is also possible for artificial intelligence.

      1. Matt Fulkerson says:

        I have some loose philosophical arguments regarding how ethical collaboration may win if you’d like to take the discussion off line some time.