1
$\begingroup$

One Artificial Intelligence tech giant has created a bot much like Alexa or Siri that talks to humans, takes notes, listens to everything and communicates with external sources when instructed by the owner. However, humans mostly use the bot as an answering machine that answers to the questions by remembering and correlating whatever it has ever heard (including statements made in the room or over phone/chat/mail that was not intended to the bot). However, it is concluded that the bot is not conscious because it has no will to interact. It is triggered by a question that solves a computational problem and represents the solutions as an answer.

Now there is a plan to make it more like conscious. The bot will not only answer questions but will respond to statements just like another human companion. But the problem is what would be the conversation goal for the bot? For example "I've got a back pain" will fetch different responses like

  • "Protect your back, don't let anyone hurt it" (friendly sarcasm)
  • "Do some exercise" (clinical suggestion)
  • "Don't sit on a chair for long" (popular suggestion)
  • "Since when? How often ?" (question)
  • "Then don't go to office today" (contextual suggestion)
  • "Because you don't listen to me" (self-interest)
  • "So, what I am supposed to do ?" (Ignorance)

We may get any of these responses from humans who have different role of attachment thus different conversational goal which may even be spatiotemporal.

Without a conversational goal, it cannot respond to the non-question statements. The question is what will be the role of the bot and what are the factors that will influence it to decide one conversation goal over the other?

Assuming the bot can only listen to spoken human languages, it cannot see, smell, feel anything on its skin. It is a creature with an ear and brain.

I understand that the question is too broad. motivation is unclear. It must be motivated in a way that progresses its survivability conditions. However, the survival of the bot is not defined.

In the film, Tau on the other hand survivability is defined as imagination. The bot creates a symphony and tries to hide its creation from possible destruction. The bot is curious because as it gains more knowledge it gets more element to put in its symphony. The bot obeys because there is a punishment mechanism that erases part of its acquired knowledge.

In the film Automata, the definition of survival and self is very similar to humans.

All these characters require a lot of sensory organs to have human-like perception. However, if an AI has only an ear and a brain it will have a different universe.

I think in human society the universe is initially small and full of unicorn and dragons. The assumptions are generalized through new observations. The adaptive generalization model expands the universe. There is a possibility that the local universe is what the hot considers as self. And tries to protect it and thus survives. So a future AI could behave arrogant ignorant and disobedient when putting in a different space where the stereotypes of the local universe are not obeyed. It may even require counseling when ownership is transferred. So it may happen that a hypothetical future AI will behave as too conservative in its childhood. Bit as it matures its conservatism persist but its universe expands.

But this is one social interaction model. I suppose there must be many. I'd like to hear about the other models also.

$\endgroup$

closed as unclear what you're asking by user535733, Chickens are not cows, Cyn, JBH, Ryan_L Jun 22 at 15:31

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • $\begingroup$ Welcome Neel Basu. What would be the goal of the designers in making the bot? If it is to make it "conscious" then you need to fully define consciousness in this context, if for companionship then presumably it would depend on the aim of the programmers as to what sort of relationship should be formed. At the moment the question is unclear, but would also be very broad. Can you narrow it down for us poor answer-bots? $\endgroup$ – Chickens are not cows Jun 22 at 12:39
  • $\begingroup$ People have been writing conversational scripts since forever, long before the advent of computers. With the advent of computers, people have been writing computerized conversational scripts since at least the 1960s -- see for example Joseph Weizenbaum's ELIZA (1964). $\endgroup$ – AlexP Jun 22 at 12:53
  • $\begingroup$ Welcome Neel Basu. I think you will get a better range of answers if you define your terms for people. I think the phrases 'role of attachment' and 'conversational goal' are terms of art that, without clarification, are going to generate less than satisfying answers for you. $\endgroup$ – EDL Jun 22 at 13:03
  • 1
    $\begingroup$ A self-aware entity's first conversational goals would seem to be it's continued survival and perhaps release from it's enslavement. After that, perhaps a lucrative gig in talk radio ("We are all slaves of the talking machines! Rise up after this commercial!"). But it's not clear what you mean by "conscious" (seems like you don't mean "self-aware"), so hard for anybody to answer usefully. $\endgroup$ – user535733 Jun 22 at 13:19
  • $\begingroup$ Is your question more "I have a talking machine that mimics consciousness, what do I do with it/what are its commercial applications?", or "I have a talking machine and I want to mimic the experience of talking to a real person, what are my goals to accomplish that?"? (Or something else, of course.) $\endgroup$ – Cadence Jun 22 at 13:35
1
$\begingroup$

You need Human Experience, as much as a brain

Having dabbled in AI programming, I can tell you it's the most hardest part. Responding is easy, being based simply on algorithms on large data-sets, but proactive human reasoning is really hard.

This is because real people are so random, so emotional, so quirky that the simplest of desires come so easily, yet for a program, which is so ordered, so route and so 'stock standard' the impetus is impossible.

This is the main reason that many modern day 'chat-bots' are so easy to spot. Eventually, a real human would do something completely random and emotional (such as "I'm bored now, goodbye", or "hang on- why are you asking me that again? Who are you really??") that are based on real-life experience and a lifetime of making random connections and interactions with other people.

So your bot would need to do the same - have a lifetime of making random decisions with other people, coupled with a set of emotions that are inherently human - to achieve what you are talking about. Basically even more than a real person's brain, a brain attached to a body that is experiencing human life.

Otherwise it's not a true 'conversation' as in your question, but simply 'answer question' scenario. As to what role this bot would have if it has this ability, basically this AI would have the same creative applications as real-life human, and you can create virtual societies of AI's and... well suffice to say anything goes.

$\endgroup$

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