I think I disagree about the hardest step being recursive self-improvement, but at the very least this process seems much more likely to go well if we can at least build human-level artificial agents that are aligned before recursive self-improvement.
I think I disagree about the hardest step being recursive self-improvement, but at the very least this process seems much more likely to go well if we can at least build human-level artificial agents that are aligned before recursive self-improvement.