AI and Sociality

One thing naysayers of AI often repeat is that there is no way that machines could never replace humans in a care role or in some sort of role that is highly involving emotions. There’s a few ways to handle this kind of claim:

  1. Undermine the actual merit of humans doing these role currently, e.g. we’re not doing as well as we think.
  2. Concede that this may be true, but humans may still need a supporting AI in any case.
  3. Argue against it, saying the future of AI holds much promise even in care-giving roles or otherwise emotional roles.

There’s probably other ways to dispute or otherwise think about this claim about AI. I’m perhaps most sympathetic to the 2nd claim. I similarly struggle as Andrew McAfee does in this video on Big Think about how AI could possibly replace, to use his example, roles like a soccer coach:

So there are a lot of jobs out there that tap into, that makes use of, that try to harness those social drives.

And one of my favorite examples of a job that we don’t think of as this incredibly elite job or this incredibly prestigious job but a job that is very unlikely, I believe, to be replaced by technology anytime soon is just a girl’s soccer coach.

And that girl’s soccer coach may or may not be a strategic genius about the game of soccer, but what that person can do, if they’re any good at their job, they can motivate a group of girls to come together to overcome rivalries and jealousies and different kinds of pettiness and play together as a team.

They can teach the value of some of those social drives like solidarity. They can help some girls who are natural leaders but might be going through a difficult period in their lives get past that and assume the roles that they’re going to be good at. They can just deal in this incredibly rich mix of social things that are going on.

Let’s say we could build a computer that could figure out all the different social things that are happening among a group of twenty-five 12-year-old girls. I think that computer is actually a long way off, but let’s say we can even build that computer. Would that computer, would that robot be able to motivate those girls, draw them together, tease out what each is really good at, get them to overcome fatigue and self-doubt and all these things, realize if they were having problems in the rest of their lives and how to help them through that?

Again, one thing I’ve learned with technology is “Never say never.” That automatic soccer coach feels like it’s a long, long, long way away from me.

I don’t have the answer to McAfee’s questions but I think it’s instructive to ask: Given the worst case scenario for AI, what could AI do to supplant these roles? I think it can be helpful to think of emotional conversations as a sort of neutral tree. Often people, when they need to feel better about their situations just need someone who understands. But “understanding” is a complex and very subjective process and often relies on (for starters) not having wires in your brain.

On the other hand, humans do have wires, we have a brain full of “wires” that connect us and our emotional thoughts. It’s possible, though I’m not sure how likely, that if a machine could mimic the ways the human brain works (something still woefully misunderstood) then this wouldn’t be so farfetched. If a machine could mimic the way humans are generally “expected” to responds to emotional situations and have some sort of voice with various tones, it could work.

I do have some sympathies to the first response as I feel like there are plenty of things humans (and I’m no exception) could do better when it comes to understanding others and trying to emote “properly”. Even if the soccer coach is better equipped to understand human emotions this nearness and familiarity is not always a good thing. Sometimes our emotions become so visceral and overwhelming to us that we let it overcome ourselves.

Humans have impulses and needs that they sometimes put above whatever actually makes sense at the time. So, for example, the coach might not actually be as good as she needs to be. But she may not have children of her own and see the soccer team as a sort of “surrogate” or way to replace that very real (and valid) need of hers. This could prevent her from stopping when perhaps it’s an unhealthy choice for her or the team she is coaching.

Either way, AI can be helpful in situations where emotions and tensions run high. In these cases they don’t necessarily have to rely on the brain wiring of human beings (something very far away, if ever). All the machines would need is a basic understanding of the sport, calculations about the different teams and what makes the most sense, so that they can win.

Of course I don’t think such a process would be perfect, but it seems to me like this would be helpful in situations where coaches might otherwise be overwhelmed with the choices they must make. And this is especially the case when human emotion becomes more and more prevalent.

Abstracting away from soccer coaches it’s also useful to note that there are robot caregivers (though they are not common) and there are plenty of AI functions in medicine that help doctors with difficult procedures. So, again, it seems to me like the 2nd response is the best one, until we know for certain we can get robots to replicate our brain.

In the case that this becomes true, arguments like 3. would have much more weight and validity to them. I’m no expert on AI or machine learning, etc. but I think the proponents of 3. have a lot to prove that I do not envy. The road to getting machines to some sort of emotional plateau seems far away indeed.

But, as McAfee said, “Never say never.”

My Patreon can be found here if you enjoy my work.

Leave a Reply

Your email address will not be published. Required fields are marked *