If some claim they are doing Artificial Intelligence by providing a Voice Assistant, what shall we expect from this Voice Assistant for the claim of doing "Artificial Intelligence" to be correct?

Examples of potential qualifications:

  • is replying "Hello" to "Hello" enough?
  • is repeating any word you said enough?
  • is identifying you by your voice enough?
  • shall it support corrections? "what is my favorite color?"/"red!"/"no, it's blue. What is my favorite color?"/"blue!"
  • shall it support emotions? "I would offer you better service if you could say thank you sometimes"
  • shall it support self-awareness? "Dear master, may I ask you a question about me?"
  • etc.

Usually, this sort of encoded 1to1 response is not considered to be AI or ML. This is because a chatbot that simply responds to a designated string with a response string is hard programmed(for the most part). A lot of companies/individuals have tried to package this as true AI but it is not.

One of the simpler examples of an AI chatbot system is a model that does some sort of tree-search or similar for the input. This is more often than not, extremely naive but would be an example of a true AI system.

So to more concisely answer your question, the categorization as an AI chatbot would have less to do with the responses, and more to do with how the model selects/comes up with those responses.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.