You know how Google’s new feature called AI Overviews is prone to spitting out wildly incorrect answers to search queries? In one instance, AI Overviews told a user to use glue on pizza to make sure the cheese won’t slide off (pssst…please don’t do this.)
Well, according to an interview at The Vergewith Google CEO Sundar Pichai published earlier this week, just before criticism of the outputs really took off, these “hallucinations” are an “inherent feature” of AI large language models (LLM), which is what drives AI Overviews, and this feature “is still an unsolved problem.”
The “solution” is to curate things, invest massive human resources in it, and ultimately still gets accused of tailoring the results and censoring stuff.
Let’s put that toy back in the toy box, and keep it at the few things it can do well instead of trying to fix every non-broken things with it.
Hilariously, Google actually used to do this: they had a database called the “knowledge graph” that slowly accumulated verified information and relationships between commonly-queried entities, producing an excellent corpus of reliable, easy-to-find information about a large number of common topics.
Then they decided having people curate things was too expensive and gave up on it.