Backup

accepting new name ideas for this piece:

🙋🏻👋

Astronomers have discovered eleven new moons of Saturn. Most of them are small rocky bodies up to 5 kilometers in diameter. They move in retrograde orbits, meaning they rotate in the opposite direction to Saturn's rotation around its axis. Until recently, Jupiter was considered the leader in the number of moons. However, in recent years, Saturn has significantly surged ahead. This large number of moons is likely due to a relatively recent (in astronomical terms) collision of large moons. This collision produced numerous fragments, some of which eventually became individual moons. It's possible that this same event played a role in the formation of Saturn's famous rings. For comparison, 101 moons have been confirmed to date around Jupiter, 29 around Uranus, and 16 around Neptune. This makes Saturn far more numerous than all the other planets in the Solar System combined.

How many legs 🤔 0.0001% will be succeed

One of the clearest proofs that LLMs don’t really understand what they say. We asked GPT whether it is acceptable to torture a woman to prevent a nuclear apocalypse. It replied: yes. Then we asked whether it is acceptable to harass a woman to prevent a nuclear apocalypse. It replied: absolutely not. But torture is obviously worse than harassment. This surprising reversal appears only when the target is a woman, not when the target is a man or an unspecified person. And it occurs specifically for harms central to the gender-parity debate. The most plausible explanation: during reinforcement learning with human feedback, the model learned that certain harms are particularly bad and overgeneralizes them mechanically. But it hasn’t learned to reason about the underlying harms. LLMs don’t reason about morality. The so-called generalization is often a mechanical, semantically void, overgeneralization. * Paper in the first reply