Systems and their Mechanisms A system's mechanisms are fast or slow. Fast mechanisms are inherent and thus adopted by default. Slow mechanisms are formed by actors and take time, energy and cooperation. Actors are things which knowingly or unknowingly take advantage of fast mechanisms and are composed of slow mechanisms....
The purpose of this post is to briefly talk about two categories of alignment and how I think AI safety evaluations should use them. I only consider responsible deployment in 20XX because I often find discussions about 2XXX to be distracting and unproductive. Also, for the remainder of the post,...
A system's mechanisms are fast or slow. Fast mechanisms are inherent and thus adopted by default. Slow mechanisms are formed by actors and take time, energy and cooperation. Actors are things which knowingly or unknowingly take advantage of fast mechanisms and are composed of slow mechanisms.
Fast mechanisms are fast because they are static in the system. Some systems heavily comprised of fast mechanisms are nature, unregulated markets and pure democracies. All of these systems can thrive under the correct circumstances, but their inherent problems are insurmountable. Natural selection is a fine mechanism for change if you're willing to wait for millions of years. Unregulated markets are... (read 731 more words →)
The purpose of this post is to briefly talk about two categories of alignment and how I think AI safety evaluations should use them. I only consider responsible deployment in 20XX because I often find discussions about 2XXX to be distracting and unproductive. Also, for the remainder of the post, I will only be talking about LLMs, since these could be the 'operating systems' of future AI's (Ex. HuggingGPT). In the near future, LLMs have the highest potential for abuse, and they will be the focus here.
As Connor Leahy and possibly others say, we should be optimistic for the sole reason that we get to build the AIs of the future. A... (read 1132 more words →)