I think Google was already ahead in this LLM game within. There were news of an engineer calling something sentient and getting fired. That must be something at

I think Google was already ahead in this LLM game within. There were news of an engineer calling something sentient and getting fired. That must be something at level of GPT-4 or beyond. But issue with LLMs is that if as a user, I ask it about how inventory valuation works in context of accounting than it might say something inaccurate. To the layman it'll seem accurate but not useful because being incorrect and for the expert, they'll already know and an incorrect output is of no use to them either. So probably Google would have evaluated the product internally in accuracy, usefulness and would have rightly, at that time concluded that such a product feature is more of a liability than an asset. Disclaimer - no inside knowledge, assuming.