Monday, November 25, 2024

Google’s Head of Search Tells Employees That AI Will Keep Making Absurd Mistakes, But They’re Gonna Keep Pushing It Out

Must read

“I don’t think we should take away from this that we shouldn’t take risks.”

Still Searching

Google’s search head doesn’t want something as pesky as nonsense AI answers to keep the company from pushing out immature products.

As CNBC reports, Google head of search Liz Reid urged employees during an all-hands meeting last week not to think of issues with AI in its core product as a bad thing.

“It is important that we don’t hold back features just because there might be occasional problems,” Reid told Google staffers, “but more as we find the problems, we address them.”

Promoted to the leadership role back in March, the veteran Googler joked to employees that they may “have seen stories about putting glue on pizza, eating rocks” — references, of course, to the viral instances of Google’s AI Overview feature spitting out nonsense when asked simple queries. It wasn’t the first time since taking the search reins that Reid has had to address AI search’s propensity for bullshit, and we’d bet money it won’t be the last.

“I don’t think we should take away from this that we shouldn’t take risks,” Reid said during the meeting. “We should take them thoughtfully. We should act with urgency. When we find new problems, we should do the extensive testing but we won’t always find everything and that just means that we respond.”

Urgent Reminder

Despite those rosy remarks, it’s unclear how that urgency is being implemented.

Though the company has reportedly implemented patches and guardrails to fix the problems infecting its AI-generated search results, people were still getting told that they could make glue pizza as recently as a few days ago (though a few passes today revealed that the company seems to have manually disabled AI responses for that specific query.)

When reached by CNBC, a defensive Google spokesperson said the “vast majority” of AI Overview responses were accurate and that upon its own internal testing, the company found issues on “less than one in every 7 million unique queries on which AI Overviews appeared.”

“As we’ve said, we’re continuing to refine when and how we show AI Overviews so they’re as useful as possible,” the spox continued, “including a number of technical updates to improve response quality.”

While it does appear that AI responses are populating for fewer queries than before, it’s nevertheless quite bold to not only dismiss the evidence of AI Overview’s immaturity as “occasional problems” and seemingly put the onus of beta testing it on the public and Google staff writ large.

“Anytime you see problems, they can be small, they can be big — please file them,” Reid urged her fellow Googlers.

More on Google comings and goings: Google Hires Weight Loss Drug Exec With Zero Tech Experience as New CFOGoogle Hires Weight Loss Drug Exec With Zero Tech Experience as New CFO

Latest article