..and how much of that AI optimism (and some pessimism) is generated by the same proprietary llms? Recursive self-enbubbling?