kanzure

Karma: 25

kanzure 6 Jun 2022 15:41 UTC
−21 points
−10
on: AGI Ruin: A List of Lethalities
Our ultimate fate may be one of doom but it may also be exceedingly positive to us. The conceiving of bad conceivable outcomes is not itself able to negate conceiving of positive conceivable outcomes, nor the other way around. Doomsaying (steel-dystopianizing?) and steel-utopianizing are therefore not productive activities of man.
There has never been a guarantee of safety of our or any other lifeform’s path through the evolutionary mysts. Guaranteeing our path through the singularity to specific agreeable outcomes may not be possible even in a world where a positive singularity outcome is actually later achieved. That might even be our world for all we know. Even if it’s always possible in all possible worlds to create a guarantee of our path through the singularity and its outcome, it’s not clear to me that working on trying to make theoretical (and practical) guarantees would be better than the utility of working on other positive technology developments instead. For example, while such guarantees may be possible in all possible worlds, it may not be possible to develop such guarantees in a timely manner for them to matter. Even if guarantees are universally possible in all possible worlds, prior to, you know, actually needing them to be implemented, it may still be less optimal to focus your work on those guarantees.
Some of those positive singularity outcomes may only be achievable in worlds where specifically your followers and readers neglect the very things that you are advocating for them to spend their time on. Nobody really knows, not with any certainty.

kanzure 6 Dec 2008 0:57 UTC
0 points
on: Sustained Strong Recursion
Re: Intel and simulating microprocessors on further microprocessors.

“Simulating a hundred-million transistor chip design, using a smaller slower chip with a few gigabytes of RAM, or a clustered computer, would certainly be possible; and if stuck with 1998 hardware that’s exactly what Intel would do, and I doubt it would slow their rate of technological progress by very much.”

When you do microprocessor design there’s a subtle distinction between the simulation of the VHDL/Verilog-type information, which is basically boolean algebraic representations that are converted into the final circuits in terms of transistors etc., versus the functional testing which I know no better name of. This ‘functional testing’ is more like quality testing, where you wire up your 128-bit IO chip to testing equipment and push bits in and get stuff popped out to do formal physical verification. On 128-bit architectures this is 2^128 tests, you’re essentially traversing through the ridiculously huge state table. In practice this is infeasible to do, even in simulation (verification of all possible states), so VHDL/Verilog/RTL-type analysis is worth focusing on instead.
- Bryan

kanzure 30 Nov 2008 4:42 UTC
6 points
on: Chaotic Inversion
http://www.autistics.org/library/inertia.html

“Wanting and Doing: A common-sense model and its limitations.

In high school, I passed many hours thinking about how I wanted to be doing my homework, being frustrated with myself for not doing my homework, making elaborate plans to try to get myself to homework… and still not starting my homework. When I’ve tried to describe how this worked to others, I’ve generally been met with disbelief. “If you didn’t do it,” they say, “You must not really have wanted to.” This idea seems to function partly as a belief about how people work, but also partly as a definition—what a person wants to do is almost defined as what they end up doing. The belief-structure underlying this—our society’s common-sense explanation for what a person does and does not end up doing—seems to go something like this:

A person is a chooser. They have an array of options laid out in front of them, and they take whichever one they most want—whichever option they care most about doing. What a person does is exactly the same as what that person cares most about doing.

I don’t know how well this model works for most people, but I know this model does not work for me, or for a number of other ACs. For the purpose of this paper, I’ll call anyone for whom this model is far from working “inertial”, and I’ll call the phenomena which make it difficult or impossible for them to connect intention and action “inertia”. I’m going to try to explore what factors effect inertia in various people, and how one might structure one’s life to make inertia less of a problem. Assumed Skill Sets

To begin with, it might be useful to look at the skillsets a person would need to have, in order for what they did to be whatever they cared most about doing. A person would need, among other things, to:
- Notice they can make a choice.
- Notice what options are possible in their situation.
- Figure out how they feel about the various options.
- Bring “online” any skills which will be needed to carry out those steps (for example, if their choice requires standing up, they’ll need to bring “online” whatever motor skills are involved in standing up. If their choice involves writing an essay, they’ll have to bring “online” all the pieces of knowledge and manners of thought involved in essay-writing).
- Begin—i.e., actually start moving, in response to thought.
Since a lot of ACs are missing various neuro-typical cognitive modules, and since if any of these steps fails to work in a given situation the person will be inertial in that situation, it is perhaps not surprising that a lot of ACs are inertial. Also, since removing various skills from that list will all result in a disconnect between intention and action, but will have rather different internal dynamics, it is perhaps not surprising that the details of how the person is inertial, and of what changes make sense to address that, vary widely from person to person. ”