Red Line: AI is not human

[Bootstrap]

I didn't read the entire article before the nurse called me back, but I picked up a Time magazine that had an article about AI, and "How people should treat them". They drew a parallel between respect given to animals, and stated that "humans are not as unique as 'they' tend to think."

This is a red line for me. If we ever start giving AIs human rights, then they are no longer a tool for humans to use as we will. And no human being is liable for disasters caused by AI.

Only humans are humans.

This is one of the big questions of science fiction — what happens if artificial intelligence obtains sentience? What happens if humans do eventually create a machine that has the ability to be self-aware, introspective and to experience emotions like humans do?

In the movie Blade Runner, Harrison Ford hunts “Replicants” — artificial humans — who have illegally come to earth to force the corporation that created them to expand their lifespan beyond four years. Rachel, one of the Replicants, believes she is a human because she’s programmed with false memories. In the Star Trek: The Next Generation episode “The Measure of a Man,” Captain Picard has to defend the android Data in a tribunal to determine whether he’s the property of Starfleet and can force him to participate in a scientific experiment that could destroy him.
Some people seem to be jumping the gun based on an illusion.

I think of Captain Kirk in Star Trek vs. "2001, A Space Oddysey". In Star Trek, the computer is very much still a computer, a tool that the Captain commands. In "2001", the computer has its own objectives, theoretically what humans asked it to do, but people really aren't in command. Here's a quote about how some computer scientists have considered this problem:

AI alignment researchers have a name for the property that distinguishes these two models. Stuart Russell, in Human Compatible (2019), argues that the standard model of AI — build a machine that optimizes a fixed objective — is inherently dangerous because we can never fully specify what we actually want. A machine optimizing hard for the wrong objective will resist being turned off, because being turned off prevents it from achieving its goal. HAL is the proof of concept.

Russell's solution is to build machines that are genuinely uncertain about human preferences and therefore defer to humans for correction. He calls this property corrigibility: an AI that accepts correction and remains under human authority.

Most of the AI companies are talking about replacing workers, networks of AI agents that can make their own decisions without human intervention, etc. Most of the people I know who are actually producing results I trust are using some kind of Captain Kirk model.

You and I can't control what the AI companies advertise, but we can decide how we use AI. People who design systems can decide whether building corrigibility into our systems, and I would urge them to do so.

[Bootstrap]

Wife: it seems like there might be some similarities between this and vaccines. Some people might have ethical reasons, and some might have perceived safety reasons, and I would be uncomfortable judging someone on either side of the fence.

I think you’ve got me somewhat convinced that it would be helpful (particularly some of the safety features like staying in your lane, etc. as long as it allows you to avoid obstacles) although I doubt we’ll be able to afford it anytime soon. On the other hand, I think it would be good for our children to learn how to drive without it so that it can be an aid and not a handicap if they ever have to drive a car that doesn’t have it or it’s not working.
As I said, it’s not like we’re going to get much choice because it’s way out of our price range.

A bit of a difference is that vaccines are a technological change to the natural order of things, and I don't really see much obligation of someone to get one. Driving cars without FSD, on the other hand, isn't exactly the natural order of things and is certainly a choice - particularly for Anabaptists.

My ethical quandary was limited to someone who could choose between, say, buying a Tesla Model Y with FSD or buying a similar new SUV, or choosing between taking a Waymo and an Uber because they live in a place like Phoenix.

I"m trying to relate this to the subject of this thread. I think there are two places where they overlap.

1. In an accident, who is liable? My wife and I rented a Mach-E for a week because it wasn't much more and we thought it would be fun. We saw the parallel parking feature, and it probably could park better than I can, but I had never used it, and what if something went wrong? Currently, the "driver" of the vehicle is liable pretty much everywhere, unless there's a clear defect in their software, so I would have an accident on my record. I think that's the way it should be, some human needs to bear the responsibility if something goes wrong. Some people suggest the AI should be a legal entity, like corporations are. That idea terrifies me, especially when you think of the scale of disasters that AI could cause, leaving all humans off the hook.

2. People don't flourish if they are passive recipients of everything. We need to drive, to steer, to think, to create, to do these things together in fellowship. I do NOT mean to say that everyone should be literally driving - seniors do need to give up the keys sometimes, some people are simply unsafe drivers or have drug problems or alcohol problems or whatever - but we need to have some areas in our life where we are in control, useful, creative.

AI companies talk a lot about "human flourishing", where they say they want to give the right answers that lead to people living well. But if there is no place for humans to be the humans, then humans will not flourish. Sometimes they talk about creating a guaranteed minimum income with all the wealth they claim AI will create. So far, they show no eagerness to even pay taxes, but beyond that, I don't think that would be a fulfilling life. What would we do all day to have meaningful lives?

[Bootstrap]

Wife: I’ve thought this before and reading the AI morality thread, it’s very fascinating how we have worked so hard to create an artificial intelligence molded in our image. Any illusion of sentience is just a testimony to how well some software developers created it.

We are created in the image of God. Unlike Him, we cannot create life, nor can we create something out of nothing. if anything speaks of intelligent design vs random chance in our universe, it is the computer and stuff like artificial intelligence.

Exactly.

You point to two really important category errors. God is God, and we are not. Whenever we treat ourselves or others as though we were gods, things go badly wrong.

Humans are human, AI is not. If we don't keep that straight, things will go badly wrong.