Meta’s decision to train his AI Patterns on social media posts from its Facebook and Instagram sites raise ethical questions about the data ingested by AI, warned Emma Christy, an analyst at GlobalData.
In a blog post, Meta explained that it would begin training its AI models to recognize colloquial language and local references from public social media posts posted on its sites by European users.
The social media giant said its AI models would not be trained on personal data, but rather to recognize speech patterns frequently used in the online sphere.
Meta confirmed that it will not train its AI models on private messages or data from users under the age of 18.
The company said training its AI models on Europeans’ social media posts would help its AI serve European users more accurately, allowing it to understand linguistic and cultural differences between European countries.
Meta said in a blog post that it was not the first tech company to train an AI model this way, saying both Google and OpenAI had also trained their AI models on European user data.
Access the most comprehensive company profiles on the market, powered by GlobalData. Save hours of research. Gain a competitive advantage.
Company Profile – Free Sample
THANKS!
Your download email will arrive shortly
We are confident in the unique quality of our business profiles. However, we want you to make the decision that is best for your business, which is why we are offering a free sample that you can download by submitting the form below.
By GlobalData
GlobalData analyst Emma Christy said Meta’s decision to train its AI on social media user data still posed important ethical questions around privacy and consent, which other companies have or not used similar training data.
Christy particularly focused on Meta’s decision to use an opt-out system for its social media users rather than an opt-in system.
“After adopting an opt-out system, Meta users may be unaware of how their data is used, which creates challenges for maintaining transparency and accountability,” she said.
“Instead of asking users for their consent (opt-in), Meta argues that it has a legitimate interest which overrides the fundamental right of European users to data protection and privacy,” he said. she adds.
Christy said relying on opt-out could shift responsibility for consent onto its users. “Users must detail their personal reasons in an objection form accessible only via several clicks and an additional login. Business implications for Meta could include increased regulatory scrutiny and decreased consumer confidence,” she added.
