IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

UNESCO: Reassert Public Control Over Generative AI

Recently addressing the disruption ChatGPT and other tools have brought to global education, the international cooperative agency recommends new laws and regulations, training and forward-thinking public debate.

An arm of the United Nations urges the world's governments to regulate generative artificial intelligence in schools.

The United Nations Educational, Scientific and Cultural Organization (UNESCO) on Sept. 7 released a 48-page report, Guidance for Generative AI in Education and Research, addressing the disruptions caused by ChatGPT and other AI tools increasingly used in schools and universities. Chief among the international agency’s grave concerns: rapid widening of the digital divide, an absence of protections for copyrighted property, and the public’s lack of control over AI uses and results.

“Generative AI can be a tremendous opportunity for human development, but it can also cause harm and prejudice," Audrey Azoulay, director general of UNESCO, said in a public statement. “It cannot be integrated into education without public engagement, and the necessary safeguards and regulations from governments.”

In a worldwide survey of more than 450 K-12 schools and universities, UNESCO found that less than 10 percent had formal policies or guidance addressing the use of generative AI applications, “largely due to the absence of national regulations,” the report said. Publishing a new textbook, it said, requires more authorizations than classroom use of ChatGPT or other tools.

According to the report, regular access to generative AI architecture and training methods is limited to only a fraction of nations across the globe — United States, China and parts of Europe — meaning that other countries, “especially in the global south,” do not have resources to create and control this rapidly emerging technology.

“As an immediate consequence, the data-poor regions have been further excluded and put at long-term risk of being colonized by the standards embedded in the GPT models,” the report said. “The current ChatGPT models are trained on data from online users which reflect the values and norms of the global north, making them inappropriate for locally relevant AI algorithms in data-poor communities in many parts of the global south or in more disadvantaged communities in the global north.”

Moreover, the report says, it is impossible to remove someone’s data from a GPT model once that AI tool has been trained, which might contravene European laws about people's "right to be forgotten" or have their data expunged.

UNESCO recommends that governments take seven steps to implement public oversight of generative AI:

  • Endorse or develop general data protection regulations (GDPRs) — Existing regulations in the European Union dating back to 2018 can be retrofitted to cover emerging technologies. Any legislation based on GDPRs should “include regular monitoring of the operations of generative AI systems,” the report said.
  • Adapt (or revise) and fund government strategies at the national level to regulate ethical use of such technology across all sectors, including education.
  • Identify principles of ethical AI use that can be translated into enforceable laws or regulations.
  • Use or update existing copyright laws to regulate content generated by AI — The U.S. Copyright Office recently ruled that “only material that is the product of human creativity” can be protected. The European Union, meanwhile, proposed regulations requiring AI developers to disclose copyrighted materials they used in building their systems. China recently passed regulations requiring the outputs of generative AI to be labeled as products of “digital synthesis,” according to the report.
  • Expand and revisit regulatory frameworks — Considering how rapidly generative AI is developing, governments need to quickly identify the gaps in local or national laws that would otherwise give them control over the emerging technology, and also consider future applications of generative AI that should be covered in new regulations. The report hails China’s labeling requirement that protects copyrighted material as an example other nations should follow.
  • Invest in AI technology and training for education and research — Much more work is needed to assure that teachers and students understand both the benefits and risks of generative AI. Training and continuous coaching for ethical AI use should be provided to researchers and educators.
  • Reflect on implications for the future — Generative AI is new to education, so it can be difficult right now to forecast consequences or rewards. “Open public debate and policy dialogs on the long-term implications should urgently be conducted,” the report said.

Overall, UNESCO endorses a human-centered approach for managing current ethical use of AI and navigating the technological journey.

“AI tools should be designed to extend or augment human intellectual abilities and social skills — and not undermine them, conflict with them or usurp them,” the report concluded.