You say you focus on technical research. Specifically, you focus in microeconomic (incl. game theory and decision theory) and computer science research. I read that you have a cognitive scientists too. Though, I have yet to see a single unit of cog-sci research output.

Cognitive architectures are generally agreed upon as the frontier to overcome to achieve human level intelligence, let alone superhuman intelligence:

Nick Bostrom asked leading researchers in the field of Ai to pick which technology provides the most promising path to human-level intelligence. The most popular answer was cognitive science. However, the second most popular answer was Integrated Cognitive Architectures.

This may not be such a problem:

• Rationality material on LW approximates a lay hypothesis generation for GOMS.

• Some reinforcement learning) cognitive architectures approximate the computer science research you do in anycase.

But that the fact that what you communicate doesn't closely match what I infer suggests something or the other is sub-optimal.