It seems fine to me. When there is evidence for a certain type of current or future harm they present it, and when there is not they express uncertainty.
Can AI enable phishing? "Research has found that between January to February 2023, there was a 135% increase in ‘novel social engineering attacks’ in a sample of email accounts (343*), which is thought to correspond to the widespread adoption of ChatGPT."
Can AIs make bioweapons? "General-purpose AI systems for biological uses do not present a clear current threat, and future threats are hard to assess and rule out."
This is like someone saying "I am much more worried about the implications of dumb humans using flintlock muskets in the near term, then I am about the theoretical threat of machine guns and nuclear weapons." Surely the potential for both misuse and mistakes goes up the more powerful the technology gets.
> Vyxal aims to bridge the gap between simplicity and "golfability".
With code golfing languages, there are inherent tradeoffs between code size and usability/fun. IMO the most important features for minimizing length (assuming the only rule is the interpreter must be published before the challenge) are:
* (in Vyxal) Efficient syntax. Not sure what state of the art is anymore but stack-based seems reasonable.
* (in Vyxal) String compression
* (not in Vyxal) Efficient encoding; Huffman coding at a minimum but ideally arithmetic coding using sophisticated machine learning to predict the next command. It's super inefficient to have each command be 1 or 2 bytes regardless of frequency.
* (not in Vyxal) Huge numbers of builtins; Vyxal has "only" ~560. Ideally every past code golf question and every OEIS sequence are their own builtin.
Vyxal might hit a sweet spot, but I'm skeptical that it actually score as well as other languages with more of these features.
Can AI enable phishing? "Research has found that between January to February 2023, there was a 135% increase in ‘novel social engineering attacks’ in a sample of email accounts (343*), which is thought to correspond to the widespread adoption of ChatGPT."
Can AIs make bioweapons? "General-purpose AI systems for biological uses do not present a clear current threat, and future threats are hard to assess and rule out."