I am quite a fan of the AI 2027 project. Its great to raise awareness of that AI could appear soon and could go bad. Yet its importance to stay sober that these “models” are really just fancy mathematical justifications for the forecaster’s intuitions1.
For reference my own perspective is that AI will more less show up in 10-100 years and that it has a high likelihood (>50%) of being very bad for humanity.
Footnotes
Given these forecasters do have a pretty good track record of guessing things well↩︎