2
$\begingroup$

People often value their pets, livestock, plants, yachts, cars, and plants.

  • Dogs are considered family members by many families.
  • Cats that might contribute little more than a purr or a look are also.
  • Horses and people develop a connection and the horse permits riding.
  • People have favorite trees and keep flower gardens and lawns.
  • Water vessels, stuffed animals, and favorite livestock are given names.

The requirement in society for an organism or inanimate object to be lovable can be quite low. The need for creating artificial pets is high for regions or housing situations were pets can't be.

What makes a dog loving and lovable? What are the actual behavioral characteristics that make it work for so many people, some of whom consider their dog a better friend than their spouse, parents, or work associates?

  • Warmth to the touch (being a mammal)
  • Fuzziness
  • Smoothness in contour
  • Aesthetic design (based mostly on millennia of breeding)
  • Attentiveness (based mostly on millennia of breeding)
  • Appreciative response to provision of the most basic things like trivially prepared food, a dish of water, or access to the yard to expel waste
  • Learned responses to positive and negative verbal expressions and body language
  • Appreciation of simple gestures, such as petting
  • Barking when someone is in the yard unexpectedly
  • Performance of the the most basic tasks, such as a golden retriever on a hunt or merely fetching a stick or a toy
  • Perhaps attacking an attacker for some of the more fiercely bred breeds

What of these is outside the realm of existing AI technology? Perhaps only the last two items, and that limitation may be overcome within the next ten or twenty years.

Although it would be nice to have a robot that cleaned the kitchen, painted the house, drove the car, or mowed the lawn, perhaps love, even if simulated, may be less superficial and more valuable than what makes some pets, plants, water vessels, and stuffed animals so dear.

How can some artificial loving pets be designed?

Is there a good architecture or algorithm for lovable artificial pets?

$\endgroup$
  • 2
    $\begingroup$ I think, that in your list, you are only talking about positive aspects of a pet, for which we love them. But I think, that in reality, even some negative aspects can make a pet loveable. For example, I hate it when my dog makes a mess of the courtyard, but this thing makes him unique, and differentiates him from example a good behaving dog, and this uniquness is one of the things that I love in him. If we say, that the uniquness, or his personality is one thing, that makes a dog lovable, that's one more thing that is still unachiavable with the current AI technology, or am I mistaken? $\endgroup$ – Adam Baranyai Jan 25 at 12:07
2
$\begingroup$

You probably remember Tamagotchies (or however thats written). They were loved by their owners but were completely artificial. Maybe a revival of those but with an AI could change the way of thinking again and already today accomplish what you suggest. A little device, maybe your phone maybe not, could bring you a little friend. It might be interesting. The only thing I personally feel: I only can consider something loveable when I can touch it an feel its warmth. My smartphone will never ever do this. Further, I think it has to be possible to die to be loved, at least this a tamagotchie can "do".

Edit: I also believe uprising and age play a major role in what can be loved and what not

$\endgroup$
2
+25
$\begingroup$

I think an AI could, to be lovable, cover some basic human psychological needs.

Following your analogy with pets, even if some pets do NOT provide most of the comforts you describe them with, they may still find a place in a human home, because they provide psychological relief. A good example of this may be parrots: none I have ever met would be happy with getting the physical treatment dogs get, they are not that thankful when they get food or water (thy just go at it), you don't get to feel their warmth all that often and so on, but they are still loved pets, because owners don't feel lonely when they are around and make noises, and because owners also feel needed.

I also think the quest for AI-based pet substitutes, or even human friend substitutes started several years ago. Just like @kishmet said, Tamagotchis have been around for a while, highly successful Pokemon games let us entertain friendships with 6 digital superpowered pets at a time, Alexa and Siri are capable of making us laugh (sort of) and perfect chat bots and sex dolls are being actively developed.

Modern media has also explored the possibility of an AI-based pet or friend with different levels of success. Examples of this are Short Circuit, KITT (Knight Rider) and Samantha (Her), all of them covering our need for friendship and safety, thus arguably making (many of) us (sometimes) wish we had such friends in our lives. AI-based villains have always been in vogue, but that's another story completely.

So that's my theory: human psychological needs covering. All other factors are only extras.

$\endgroup$
1
$\begingroup$

Out of cats, dogs, plants, parrots, snakes, yachts, cars, stuffed animals, and other things people give names to (indicating some psychological attachment), dogs are probably the most often included in a holiday photo of the family. Maybe that's because cats won't pose, but that only furthers the notion that dogs occupy the most valued place in the lives of families. They are also the most disliked by housing associations and landlords, so the product idea is a good one.

The mechanical, materials, manufacturing, and aesthetics experts can handle the temperature, fuzz, smooth, and sleek part. The AI must make the a-pet attentive, appreciative, responsive, and able to detect the sound an intruder might make. We can leave the fetching birds and sticks, attacking intruders, and friendly licking with a wet tongue for when that level of robotics is technically feasible and inexpensive enough for families. Since NLP isn't really in a place where the AI can be counted on to get things like encouragement right, except in a fortune cookie kid of way, so that's better left for a future model too.

The a-pet needs and a high level algorithm like this.

  1. Run through product initialization with the owner, which includes the acquisition of samples of its own name as spoken by members of the family, along with ways the family and especially the main owner will express to the e-pet when it is behaving well or not so well

  2. Recognize its name when spoken by family members

  3. Detect the relative angle of the source from the phase and magnitude differences between the left and right channel like people and dogs do

  4. Turn to find the human face in that direction

  5. Identify the human face that probably spoke

  6. Recognize the facial affect and gestures

  7. React in a way that will please the owner, which may include things like hyperactivity, tail wagging, barking, purring, smiling, growling, moving toward the face in a friendly way, showing other forms of friendliness through movement or sound, and backing up slightly and taking a stand as if to pounce if the face is not familial and shows some predatory indication

Since speech synthesis sounds much more natural this year, the reactions may include friendly and generally encouraging things, but as said above, waiting for a deeper comprehension to support real encouragement might be a better idea.

The reaction will need to be guided by owner interests, which means that petting or some kind of signal in speech needs to be trained in when the e-pet is taken out of the packaging and initialized or somehow trained in at the factory. That's part of step one in the algorithm. That becomes the value function for reinforcement or some other kind of continuous, real time learning AI system.

Every one of the seven steps seems feasible with today's AI technology. The e-pet can be made. Will the price point meet the market interest? Expensive dogs are purchased every day. I think it might fly.

$\endgroup$

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.