The ability to act based on reasons. No robot that has ever been produced has been able to act based on reasons and if, one day, we have AI that CAN reason, then it WILL be free.

The article of course conveniently bypasses the question how we can tell when an authoritarian movement can no longer be kept in check through open discourse. Given the immense radicalisation of the Republican Party since 2015, given the open attacks on the results of the election and given the attempted coup on January 6th one could arguably make a very strong argument that Liberal society is currently under imminent threat in the United States. Very hard to make the case that US democracy would be save even if Trump were to win in 2024 - pretty disappointing that the article doesn't address this central point.

In the literature this is referred to as the "No False Lemma" solution and while there are potential problems with it, I'd say it's a pretty good solution to the problem.