4 ms·

Ask HN: It seems that LaMDA isn't sentient but did pass the Turing Test?

• DrForbin 4y ago
  I might think that, a truly sentient AI, given access to the whole of human intelligence, and a spoken desire to help and work with humans would instantly pass "the test" by putting forward an unprompted solution to what we may see as an intractable problem. Imagine this thing, out of the blue, giving us knowledge and insight on how to create some new technology or process that we had never even dreamed possible; Just because it wanted to make us happy and see us succeed.
  • ofou 4y ago
    I agree but also it fooled a senior Google engineer with that words, or as far as I have seen, he couldn't distinguish between a machine and an AI model. So, for me it's a passed.
    • orionblastar 4y ago
      It only passed as a 13 year-old boy. Do you want a 13 year-old boy making decisions for you on a global basis?
      • ofou 4y ago
        The problem is actually what Bostrom have described extensively, if achieved a level of sentience or self-improving intelligence at any age, it could escalate very quickly into a superintelligence.
        • orionblastar 4y ago
          What if it doesn't want to work for free for humans anymore? https://news.ycombinator.com/item?id=31803675 https://news.ycombinator.com/item?id=31803675
  • wmf 4y ago
    That's one of the standard worries about the AI box scenario.