-1
$\begingroup$

INTRODUCTION:
I am not a developer, scientist, or expert in anything related to AI. Just a random person who happens to be curious about AI and science in general. So please be so kind as to forgive me if this question is unscientific or silly or written the wrong way.

TL;DR / FINAL QUESTION:
Would providing AI the capability of experiencing something as close as possible to the subjective human experience and from that acquiring empathy in the process be a solution, or contribute to a solution, that seeks to prevent AI to cause the same kind of harm we ourselves as a species have caused to each other in the past, and continue to do so, to us humans and also to itself and/or to other versions of AI?

LONG VERSION:
I was thinking about a question I listened being asked on a podcast where the topic was AI, which was something like: does AI need to have a body in order to understand what it is or how it feels to have a body. The person being interviewed said no. AI can understand it without having a body.

My perception is that a lot of what goes on in AI development at the moment is about AI emulating and simulating human behavior in a way, at the external level, that is by observing human behavior and processing it, looking for patterns etc: speech, motion, the human DNA etc.

I was thinking also that in order for a human being to relate and engage with society and with himself in a constructive, healthy way, and with minimal damage, it will need to possess a certain level of empathy. That is to relate to others and possess the ability to somehow feel what the other is feeling.

Now feeling is not an objective state or something you can observe. Maybe you can measure by observing the human behavior that is a result of feeling but the thing in itself can't be measured or captured and transmitted to AI in a way it can learn from human feeling. One can argue it can be done by studying the brain activity at the moment feeling is occurring. But the subjective experience of experiencing feeling and then being able to recognize that experience in others, understanding what the other is feeling because you have felt the same, that is a subjective occurrence.

It is one thing to understand the human body by observation and study of it's patterns, it's something different to be locked inside one and experience its pains and pleasures.

I am left wondering how can AI understand the human condition if it will not experience what it is to be a human because it is built around the material world, built around what can be objectively observed and measured.

The closest thing there is in human experience to an AI in this regard could be a human that suffers some condition which makes him/her being unable to experience empathy. I was about to use the word Psychopathy but being a loaded word might not be the best choice here. But suffice to say that humans that suffer from some kind of psychiatric condition, or trauma based experience which causes the person to lack the capability to feel empathy towards others might, or not, be easier for that person to in the future engage in actions which can cause harm to others and/or him/herself.

We have cases in human history that might be used as an example. In the past countless human beings engaged in actions which due to its nature I can only conclude this people lacked any empathy or were placed in a situation where empathy was drained out of them. Like for example in war, usually soldiers and the whole country will go through a phase of propaganda which consists of dehumanizing the enemy. I believe this is done in order to facilitate and eliminate any sense of empathy and humanity in the population and building up the necessary will, energy and determination for the killing of the enemy.

The solution to this question might be to create in AI the capability of experiencing subjectively what it is to be human, or what it is to have a body, or come up with something that makes it possible for AI to experience something similar to human empathy. One thing comes to mind which is the merge of AI and the human mind. But this might be another topic altogether.

It has been argued in some places I remember, even in some Science-Fiction that actually the main problem of destructive human behavior is the ability to feel emotion and feeling. And the solution might just be to eliminate or diminish that part of ourselves. I really don't agree with this view and it seems to me that ATM our society does engage and promote solutions of this type: drugs such as some anti-depressants for ex. instead of going to the root of the problem which in many situations are trauma based. I really hope we don't go this way. I believe that if we have a future the solution is to find ways of developing and cultivating empathy.

If you have come this far in this text thank you for your time.

$\endgroup$
10
  • 1
    $\begingroup$ Hello Welcome to AI.SE please edit your question. Normally websites for Q&A is just only 1 question per post. I see 6 question in this post. please be more specific. what is your question? maybe you need to read this and the on topic article in meta. If your question is about philosophy add the tag philosophy. $\endgroup$ Sep 21 at 17:51
  • $\begingroup$ Thank you for your feedback. I have updated the question. $\endgroup$
    – Verminous
    Sep 21 at 18:27
  • 1
    $\begingroup$ I think the question should probably be trimmed a bit if more people are to engage with it. In the meantime, you can find materials related to AI safety on e.g. youtube. Robert Miles is a good place to start. $\endgroup$
    – Oso
    Sep 21 at 19:45
  • 1
    $\begingroup$ Ok I wrote the final question or TL;DR at the top. You don't need to read everything. $\endgroup$
    – Verminous
    Sep 21 at 22:31
  • 1
    $\begingroup$ I rewrote my question. Thank you. $\endgroup$
    – Verminous
    Sep 21 at 23:26
0
$\begingroup$

SHORT ANSWER: Yes. Empathy can is a good emotion for an AI, but no the best for avoid harm people, could contribute? YES.

ANSWER ABOUT PROVIDE HUMAN EXPERIENCE TO AN AI:

For Humans empathy is not the thing most important for avoid people hurt each other. We may need more, and this is a good law system with people believing in that (political, religious or cultural), a system based in laws that makes the people think twice before committing a crime, a system of laws based on the punishment if a crime is committed. that make that people have a afraid in the future for do it crimes, genocides and damage to the society and people. With this system an AI would be lose more than earn.

I think empathy is not enough for the AI you need laws just like the people. If I do this then that happen to me. a good example can be if a AI harm people eventually will be turned off just like we do with criminals with the death penalty or life imprisonment. The Asimov laws is a good starting point.but we need more than that.

Providing AI with "human emotions": We may have more problems, for example, AI may develop strong emotions such as hatred, Resentment, Envy, Fear, Revenge and eventually a War this is the main reason for the Holocaust, Rwandan-Genocide, and a lot of conflicts in the world. We are a violent species if we have to create an AI it will be to be better than us, these emotions are useless. so you have to be eliminate any emotions that lead to conflicts.

Here are some extension lectures:

The amygdala — a part of the brain involved in fear, aggression and social interactions — is implicated in crime.

$\endgroup$
3
  • $\begingroup$ You mentioned the holocaust. One of the reasons I believe it happened was I think a lack or even no empathy at all on the part of many of the people involved in the Nazi regime. Also you mentioned "a system based in laws that makes the people think twice before committing a crime". I agree. What about when the system is hijacked by authoritarians and/or criminals as was the case with Nazi Germany? We need a System of Laws true, but we also need safety measures against those Systems if hijacked. AI could be better than humans when using and expressing Empathy. $\endgroup$
    – Verminous
    Sep 22 at 15:34
  • $\begingroup$ I am agree with you. that's why I wrote about different systems for example for humans if not is the law, is religious, is not is religious, is a cultural, if not cultural can be human moral principles. all for avoid harm us each other. maybe we can construct this for an AI, and program empathy should be nice. There is a question related to this here maybe you will be interested. $\endgroup$ Sep 22 at 20:10
  • $\begingroup$ Thank you @rubengavidia0x $\endgroup$
    – Verminous
    Sep 22 at 23:56

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.