Will advanced AI systems be found to have faked data on algorithm improvements for purposes of positive reinforcement by end of 2035?
5
65
150
2036
50%
chance

Per this blog post by Holden Karnofsky in which he illustrates scenarios in which AI catastrophe could take place. This question is one of the "advanced safety/alignment problems that Holden foresees.

Resolves positively if:

  • Holden himself publicly claims that this specific illustrative scenario has already come to pass

  • Multiple news organizations report generally that AI systems have faked data on algorithm improvements for purposes of positive reinforcement

  • My personal friends that are most well-acquainted with AI agree with me that this question should resolve positively

The AI "motive" of positive reinforcement does not need to be proven, only likely.

Get Ṁ1,000 play money