The post must be a public post on the Astral Codex Ten blog. To qualify as "substantial," the post must contain at least 1,000 words (including headings, titles, subtitles, image captions, and anything else that can reasonably counted as "part of the text") that are primarily about shard theory (common sense applies for assessing this as well as any other ambiguities, but I will make the final judgment).
Quoting the AI Alignment Forum:
"Shard theory is an alignment research program, about the relationship between training variables and learned values in trained Reinforcement Learning (RL) agents. It is thus an approach to progressively fleshing out a mechanistic account of human values, learned values in RL agents, and (to a lesser extent) the learned algorithms in ML generally.
Shard theory's basic ontology of RL holds that shards are contextually activated, behavior-steering computations in neural networks (biological and artificial). The circuits that implement a shard that garners reinforcement are reinforced, meaning that that shard will be more likely to trigger again in the future, when given similar cognitive inputs."
Here are all of the posts on the Astral Codex Ten blog (as far as I can tell) that covered AI safety in the past 12 months (as of mid-February, 2023):
🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ92 | |
| 2 | Ṁ64 | |
| 3 | Ṁ31 | |
| 4 | Ṁ27 | |
| 5 | Ṁ14 |