Google Image results for several search terms are showing hyper-realistic AI-generated images and graphics alongside genuine photos – without flagging the incorrect results, raising fears about misleading users and spreading false information.
For example, multiple users on X shared a graphic claiming that when looking up photos of a baby peacock on Google, they saw AI-generated images of the chicks that looked nothing like reality.
The Hindu also noticed several inaccurate images when looking up baby peacocks, as well as other baby animals. However, images of celebrities such as Elon Musk and Donald Trump yielded accurate photos, as well as place names such as Gaza and Tel Aviv.
Google is injecting more AI-generated results and summaries into its search platforms, in spite of earlier issues with accuracy and safety.
While large AI companies have committed to better watermarking techniques and identifying images made with their text-to-image tools, invisible watermarks may not be immediately noticed by users rapidly scrolling through pages of images.
Some users suggested customising Google image searches to exclude results from before 2022, to cut out the images created with generative AI. For reference, OpenAI’s ChatGPT was introduced to the public in late 2022.
However, this approach could also filter out more recent news updates and media that internet users may require.
Published – October 09, 2024 11:58 am IST