The leading AI developers are in a race against each other facing intense market and other pressures, resulting in the breakneck speed of technology development that we are currently witnessing. As I will show in the talk, this speed can be expected to increase further, possibly by orders of magnitude, as we are now entering the regime of recursive AI self-improvement. This has potentially enormous societal consequences, including existential risk on time scales that count in single-digit years rather than decades. As world citizens, we have a choice between either idly standing by watching this drama unfold, or raising our concerns.