- Meta uses public Instagram and Facebook images and text to train its AI text-to-image generator.
- Meta executive Chris Cox told Bloomberg's Tech Summit that the company “does not train on private equipment.”
- The chief product officer's comments come as big tech firms race to acquire data to train AI models.
Big tech firms are clamoring for AI training data and Meta seems to have one big advantage over its competitors: using photos from Instagram and Facebook.
Meta's chief product officer, Chris Cox, told Bloomberg's Tech Summit on Thursday that it uses publicly available images and text from the platforms to train its text-to-image generator model.
“We don't train on private things, we don't train on things that people share with their friends, we train on things that are public,” he said.
Cox added that Meta's text-to-image model can produce “really amazing quality images” because Instagram has “a lot of images of art, fashion, culture and just people and our photos.”
Users can create images on Meta AI by typing a prompt starting with the word “imagine” and it will generate four images, according to its website.
AI models need to be fed and trained on data to be effective. This has been a controversial issue for almost as long as it has existed. There is no way to stop Copyright material should be scraped from the internet and used to create LLM.
However, the US Copyright Office has been trying to deal with the problem since early last year and is considering updating its rules to deal with it.
One way companies are trying to capture data is by joining forces with other firms. For example, OpenAI has partnered with several media outlets to license its content and develop its models.
Meta also considered acquiring publisher Simon & Schuster to get more data to train its models, The New York Times reported last month.
Along with raw data sets, companies use “feedback loops” to train their models — data collected from past interactions and outputs that are analyzed to improve future performance. is done It includes algorithms that notify the AI models when a mistake is made so that it can learn from it.
Meta CEO Mark Zuckerberg told The Verge last month that feedback loops would be “more valuable” than any “upfront corpus.
Metta did not immediately respond to Business Insider's request for comment, which was made outside normal business hours.