Connect with us

Tech

Google seems to be manually removing some strange AI search answers

Published

on

Google seems to be manually removing some strange AI search answers

CEO Sundar Pichai at Google’s IO conference.
JOSH EDELSON/GETTY

  • Google is disabling some of the strange answers provided by its AI Overviews feature.
  • The company told The Verge it was “taking swift action” to remove certain queries. 
  • Google told Business Insider the strange examples were “generally very uncommon queries.”

Google appears to be manually deactivating some of the strange answers provided by its new AI search feature, The Verge reported.

The new AI Overviews have been instructing users to do odd things, including putting glue on pizza and suggesting they eat rocks.

Some of the stranger answers started disappearing after they were shared on social media, leading to speculation that Google was manually removing them.

Google spokesperson Meghann Farnsworth told The Verge the company was “taking swift action” to remove AI Overviews on certain queries “where appropriate under our content policies, and using these examples to develop broader improvements to our systems, some of which have already started to roll out.”

Google did not immediately respond to a request for comment from Business Insider, made outside normal working hours.

The company previously told BI the examples of AI Overviews shared on social media were “generally very uncommon queries and aren’t representative of most people’s experiences.”

Google said the vast majority of AI Overviews “provide high-quality information” and that “extensive testing” was conducted before the product was launched.

“Where there have been violations of our policies, we’ve taken action – and we’re also using these isolated examples as we continue to refine our systems overall,” the company said.

It’s not the first time Google has been in trouble following the launch of an AI product.

After releasing its Gemini model earlier this year, the search giant found itself at the center of a “woke” culture war. Google paused Gemini’s image-generating feature after users complained the model generated historically inaccurate images of people of color.

The chatbot also faced criticism for some of its written responses, including a widely-shared example in which it appeared to be unable to decide if Elon Musk or Adolf Hitler was worse.

The controversy caused some critics to claim that left-leaning workers have had a disproportionate influence on Google’s culture, which can, in turn, affect how AI models are built.

On February 28, Axel Springer, Business Insider’s parent company, joined 31 other media groups and filed a $2.3 billion suit against Google in Dutch court, alleging losses suffered due to the company’s advertising practices.

Continue Reading