I'm making significant progress with Rasputin as regards the concept of value judgments.
Obviously, that's something artificial intelligence isn't set up to tackle, but we need Rasputin to be able to make that leap.
Here's a hypothetical: say Rasputin has to redirect a malfunctioning warsat, and there are only two paths for it to take. One will breach the hull of a one-seat planetjumper, instantly killing its pilot. The other path will disable a Venusian pleasure craft, leaving its inhabitants alive but certain to die of starvation before they can be found.
To my surprise, when given this problem, Rasputin actually rendered a verdict.
I'm not going to tell you what it is, because I want you to enjoy your next Venusian pleasure craft trip.
Rasputin continues to develop by leaps and bounds.
In a trial simulation, he successfully interpreted the meaning of corrupted, borderline incomprehensible data from a damaged warsat. This is exactly the sort of function we need him to fill. But that wasn't the amazing part.
The mind-blower was that without proof, evidence, or conscious reasoning, Rasputin sensed which warsat was in trouble.
You heard me. Rasputin has developed intuition. I wish he could celebrate with me.