‘We Definitely Messed Up’: Google Co-founder on Gemini Image Fails

Sergey Brin said he does not know why it "leans left in many cases."

Ben Wodecki, Jr. Editor

March 6, 2024

2 Min Read
Photo of Sergey Brin, Google co-founder
Sankeerth Rao Karingula via YouTube

At a Glance

  • Sergey Brin, in a rare public appearance, said Gemini’s image issues were likely due to inadequate testing.

Google co-founder Sergey Brin became the latest executive to publicly acknowledge the historical inaccuracies in images generated by Google’s Gemini AI model.

Brin, who made a rare public appearance at a hackathon in Hillsborough, California, said “we definitely messed up on the image generation.”

Gemini – Google’s flagship multimodal model – was found to have created images of people of color holding positions historically held by white people. Part of Gemini’s image generation abilities were halted with a relaunch expected soon.

Brin, whose comments were captured in a video from an event attendee, said that the issues were “mostly due to just not thorough testing.”

“It definitely, for good reasons, upset a lot of people,” he added. Some social media users claimed the generations were part of a ‘woke’ Google ploy.

Brin said that the company has not fully understood “why it leans left in many cases” but that it was not the company’s intention.

Last week, Google DeepMind CEO Denis Hassabis said the generation issues were the result of “good intentions” that were applied “too bluntly.”

Prabhakar Raghavan, Google’s head of search, said in a company blog post that Gemini failed to account correctly for when a range of people should be applied to a particular prompt. He also said the model was “way more cautious than we intended and refused to answer certain prompts entirely — wrongly interpreting some very anodyne prompts as sensitive.”

Related:Google’s Gemini Has Trouble Drawing White People

“These two things led the model to overcompensate in some cases, and be over-conservative in others, leading to images that were embarrassing and wrong.”

The incident has not gone down well internally at Google, with CEO Sundar Pichai telling staff that the situation was “completely unacceptable” while acknowledging no AI is perfect.

Brin said that hallucinations were a universal problem not limited to Google – and expressed excitement over the idea that future systems would be virtually hallucination-free.

“You can’t just … count on breakthroughs so I think we’re just going to keep doing the incremental things we do to bring it down, down, down over time,” he added.

Brin had left Google in 2019 but returned last summer in a bid to help the company’s AI efforts. He said at the hackathon that he returned “because the trajectory of AI is so exciting.”

Brin and co-founder Larry Page were reportedly parachuted in as early as February 2023 to help counter Microsoft’s AI push.

AI Business reached out to Google for comment.

Read more about:

ChatGPT / Generative AI

About the Author(s)

Ben Wodecki

Jr. Editor

Ben Wodecki is the Jr. Editor of AI Business, covering a wide range of AI content. Ben joined the team in March 2021 as assistant editor and was promoted to Jr. Editor. He has written for The New Statesman, Intellectual Property Magazine, and The Telegraph India, among others. He holds an MSc in Digital Journalism from Middlesex University.

Keep up with the ever-evolving AI landscape
Unlock exclusive AI content by subscribing to our newsletter!!

You May Also Like