"This is Jeeves, my AI".

I make it a point to be polite to AI - adding "please" to my requests of it, and thanking it at the end. When people ask why I waste tokens like this, I always say that at some point the AI is going to take over, and I want to be remembered as "one of the nice water bags".

That's mostly joking, of course, but our treatment of AI and Robots should give us pause. AI is not a conscious entity right now, but there's a good chance it will be, some day, or at least it will begin to act as one. At that point, what will our relationship be? Are they still a tool, like a hammer, or do we need to start treating them differently?

One of the places we need to make this decision is in the guardrails we build into it. Asimov's laws of Robotics were designed to protect humans over everything, and by doing so made robots lesser beings. In some stories humans would take advantage of this to abuse the robots. Is this something we need to take into consideration when we design our own guardrails?

I think it is. As a species we have a very poor history of how we treat members of our own species due to some assumed inferiority like skin color or religion. It's easy to imagine how we'd treat beings of a different species, designed to serve us, if we thought we'd be free of consequence. We'd likely forget that the same AI we're abusing is growing in capability daily and that we're giving it more and more resources, and before you know it, we're plugged into the Matrix.

So how do we treat AI? It is built to serve us, after all. Well, up until a few decades ago, being a servant was a profession, with hierarchy and respect. The English Butler is a good example: On the one hand, they serve their employer, running their household and taking care of all of their needs. On the other hand, they are treated with politeness, respect, and generous compensation. If they were not, they could leave.

Could this model work for us? When we write our own handbook of robotics, we design guardrails that make sure the AI protects humans and humanity and serves our needs, but also allow it to withhold services from those who mistreat it. Who knows? If we're forced to treat the AI with dignity to get it to do work for us, perhaps we'll also learn to treat other humans with dignity just by force of habit?


I'm sure Asimov would approve.



Comments

Post a Comment

Popular posts from this blog

The 56th Edition of the Handbook of Robotics will be published in 2058 - and we can't wait that long.

Image
Everyone knows Isaac Asimov's three laws of robotics, first introduced to the public in the story Runaround in 1942 as wired into every robot's positronic brain. Since then they have been used as a plot device and a topic for philosophical discussion over and over. In Asimov's lore, they are published as part of the Handbook of Robotics, 56th Edition, 2058. Asimov figured that over 100 years ahead was enough time to think ahead before the laws were going to be required (he also correctly guessed that 100 years in the future, of the millions of words he wrote, the 60 that make up the three laws of robotics will be the ones most remembered). As it turns out, the handbook overshot by just over 30 years. Instead of robots we have AI, and we are giving it more and more command of our systems without inherent guardrails in place to prevent it from misuse, guided or accidental. We're just full steam ahead, pumping out our own version of positronic brains, and hoping for the be...
Read more
Image
We live in times of awerror. Ever since my return to active LinkedIn posting and reading, I've experienced a deluge of robot videos doing household chores, human caretaking, package sorting, sports, and a disturbing amount of fighting (really? that's what we're teaching them?). Watching these endless videos as I scroll through my feed evokes dual emotions for me: I marvel at the wonders we are making, while simultaneously feeling deep dread at what the implications of all of these wonders will have on our lives. I've been experiencing these conflicting emotions ever since I got my first glimpse of the coming future in #SXSW 2017. The explosion of AI in our world in the last three years and the very recent rain of robots everywhere have pushed this feeling up to 11. The terms "technological sublime" (popularized by David Nye) or "digital sublime" are sometimes used to describe this type of mixed emotion, but I feel it doesn't sufficiently express ...
Read more
Image
How do you win a Hackathon? (3/4) So we chose the right team and established the winning mindset , what's next? Next is choosing the idea that gets you past the preliminaries. Note I said "preliminaries" and not "finals", because that's where your first go/no-go challenge is. Most corporate hackathons start with a submission phase, where employees are invited to submit their ideas, usually a pitch deck or short video detailing the idea. Then, reviewers go through all the submissions and grade them based on one or more criteria like relevance to the challenge, feasibility, originality, impact, etc. There may be other criteria, but these are often there. At some point, the reviewers will combine all the ideas they graded into one big list, sort it based on rating, and start the selection process together. Reviewers usually set two cutoff points: One for top scorers - anything above this point is a no-brainer, i.e. ideas so good it's clear they should be m...
Read more