UNESCO warns that AI threatens the memory of the Holocaust

The survival of the memory of Holocaust this “threatened” by artificial intelligence (AI) generative, which frequently feeds on distorted or malicious sources to create its content, warned this Tuesday the UNESCO.

A joint report published today by UNESCO and the World Jewish Congress (WJC) points out that AI developers, by training their tools with information circulating on the Internet, can expose them, due to lack of adequate supervision, to sources with false or biased data. .

Furthermore, and on that basis, the AI ​​tools themselves can end up generating false content about the Holocaust, something that has already happened in some cases with ChatGPT and with Bard, Google’s tool.

Both have produced content about events related to the Holocaust “that never happened”, UNESCO states in a statement. Another problem encountered is that AI tends to “simplify the historical context”, privileging “a narrow range of sources and a small choice of events.”

For example, on AI-powered search pages, between 60 and 80% of the main images in searches refer to the Auschwitz-Birkenau extermination camp.

Because the 80% of young people between 10 and 24 years old use AI several times a day for educational or entertainment purposes, action is needed “fast” so that these technologies are guided by ethical principles, UNESCO insisted.

For this reason, the organization urged States and technology companies to accelerate the implementation of its recommendations on the ethics of AI, which have already been integrated into the legislation of some fifty countries.

It may interest you

• The man who survived the Holocaust and Hamas militants
• Auschwitz survivors sound alarm 75 years after liberation
• UNESCO launches observatory to monitor how countries regulate ethics in AI