We Need a Consolidated List of Bad AI Alignment Solutions

I think it might be better to have an explicit bar a potential solution must clear before it can be considered serious. Consider this quote from Sean Carroll about a potential new physical theory that would replace or subsume an existing one:

There are many phenomena that fit together; the expansion of the universe, the early universe and nucleosynthesis, the growth of structure in the universe, the fact that there are gravitational time delays as well as gravitational red shifts, there are many phenomena, all of which fit into the usual kind of picture, and if you say well, the usual kind of picture is not right, I have a new picture, you gotta re-explain all of those phenomena. Okay. So if someone says, “I have a theory about the universe that is new and exciting,” just say, “Alright, in your theory, what is the ratio of hydrogen to helium produced in the early universe?” And when they tell you that, ask them what the amount of deuterium and lithium is? Okay? Until you get to that level of detail, which is the standard model, explains very, very well, it’s not really worth taking alternative theories very seriously.

I suspect there can be a list of questions that each new alignment approach would have to answer, such as, idk, dealing with treacherous turns, out-of-distribution self-correction etc.

Now break your own proposal! It's a really useful exercise.

• Why might this be difficult to implement (remember, at the end of the day we have to write code that implements this)?
• How might this go wrong, even if it is implemented as specified?

In general I think people with alignment proposals should think for at least 5 minutes about why they might not work.

Meh.

I think the key problem with most bad ideas is that they failed to understand enough about the basic structure of the problem to know what a solution would even look like. The probability that you've hit on a useful solution without that understanding is nil, so fixing the problem requires learning more, not sharing more uninformed guesses. For example, among other things, your solution doesn't account for value loading, the question of how we get the system to have the value we care about.

We (Rob Miles and Stampy's team) might run an event where people can present their ideas for solving alignment with a much lower bar for entry than usual in a few months, once we've got a strong base of volunteer editors to absorb and filter the incoming ideas. Provisionally called "Big Wheel Of Cheese Day". If we do this we'd likely host the database on Stampy.

The AI Box:

A common idea is for the AI to be in a "box" where it can only interact through the world by talking to a human. This doesn't work for a few reasons:

The AI would be able to convince the human to let it out.

The human wouldn't know the consequences of their actions as well as the AI.

Removing capabilities from the AI is not a good plan because the point is to create a useful AI. Importantly, the AI should be able to stop all dangerous AI from being created.

I was happy to read your Golden Rule idea...I just posted my own version of the Golden AI Rule [EA · GW] a few days ago. 

The Golden AI Rule: AI will only be as good to us as we are to each other.