What will fembots like?

[Deep Blue]

What will fembots like if they are made as human-like? Will they behave like humans, will they study, work or getting old?

At first, I think they do not need to study, as they can download the data on the internet directly to their disk, they do not need to be read like mankind.

But another hand, I think fembot still need to work, the robot are made for working, if the AI do not rebellion, I think fembot still need to work.

And then, about getting old, I think fembot will not getting old, as they can transfer their data to a new robot, they can become new again, they will not getting old.

[Condor621]

I reference aging in a couple of stories.. I think as an AI develops and grows, it will reach a point that it will no longer be transferable as it will become so unique and entangled that any attempt to transfer it would fail. As for physical aging, as long as worn components are repaired and power systems are maintained, the aging process would just involve when the AI and Memory systems reach their limits of growth... kind of like how people start to forget things or adapt as they age. I could see an AI getting to the point where it just becomes a static entity and decides to not interact with the outside anymore. Like how some very old people become shut ins. Just theories...

[dubhdanaidh]

I think it depends on the A.I. That being said, as long as they’re not dominating humans I think at least some A.I. will made to be human like and then it becomes a question of how human like.

[BebesIsPerfect]

Fembots will like whatever they're told to like. That being said, if they are programmed to approximate a typical female human (whatever "typical" is), then anticipate similar likes. Results may vary.

Regarding aging, personality-wise, I don't see them getting "older and wiser" in the traditional sense. Their personality would remain steadfast if that is how they were programmed, therefore, not wiser. Their physical bodies, however, will still age and decay like Condor621 points out (therefore, older), but while our human bodies are kind of one and done, they could probably have more readily available replacement parts, assuming the parts remain In Stock. It's kind of sad to think of their life spans and livelihoods being dependent on a corporate entity. *Shouts to the back of the auditorium facetiously* Don't you think so, humans?!

[D.Olivaw]

<Multivac>
INSUFFICIENT DATA FOR MEANINGFUL ANSWER
</Multivac>

[D-wolfmaster_2022]

While robots will definitely be able to have information programmed into them, I think that they will still have the option of actually learning about topics and how to do tasks through hands on experience. Perhaps learning in the field will have the information stick better than simply downloading it to their CPU since they actually need to use said data more often in their day-to-day lives. Knowledge remains more clear the more it's utilized. They might even delete old data that's decidedly less useful to their lives; core programming excluded, of course.

[D-wolfmaster_2022]

What I'm interested in is how will the fembot feel?

How would a fembot feel knowing that it is, at best, a tool/toy that thinks? Would they be able to feel at all? Would it be able for a robot to feel emotions or at least some technological equivalent? I know I'm dipping a bit into fantasy here, but it's a still pretty interesting conundrum to ponder.

Like, let's say a fembot can have emotions. If so, how would they get them? Would they be programmed into her from day #1, or would it be something that she gains and develop over time? How would they read onto the fembot's processors? Would her CPU be able to handle emotions? Would feeling emotions cause problems for a robot? How would humans deal with a feeling thinking machine?

[dubhdanaidh]

I think it depends on how human like a fembot is or if it is even possible. I imagine likes could be randomly generated when compiling a personality. That being said, depending on whether the AI can learn it is feasible new likes and interests will emerged based on a whole host of factors.

Of course, if a fembot is nothing than a program running on a computer, there is the question of whether it can really like anything. But that raises larger questions about humans and likes and preferences. Do I like McDonald's because the food is good, because I like the taste, or because as a human I'm preprogrammed to like fats, salts, and sugar?

[silurian_brutalism]

It would depend on a lot of things. Since a gynoid/android companion would require the ability to adapt to users and new environments, they would probably be able to learn in real time, unlike the current SOTA AI models, which are static.

I disagree with the idea that they would like what they would be programmed to like, as some here postulate. Neural networks, which are most definitely going to be the basis for any embodied AI in the future, are grown, not programmed. You can do reinforcement learning, which takes a while. But also tweaking neurons and weights, which is hard and prone to messing things up. The vast majority of AIs today are black boxes. Mechanistic interpretability is an active area of research, but this will only get more difficult as the networks become more and more complex. So the best bet would be to let them learn on their own after the initial model is trained.

A gynoid will end up liking things depending on how her weights are configured inside of her neural network, similar to how you like certain things based on how your brain is wired. And tastes change as new connections are made and old ones are deleted.