Experts Alarmed by Meta's Plan to Open Source Powerful AI Before Adequate Safeguards
-
Mark Zuckerberg wants Meta to build an advanced, human-level artificial intelligence (AI) system called artificial general intelligence (AGI). This alarms experts who think it could be dangerous if misused.
-
Zuckerberg raised the possibility of making Meta's AGI system "open source." This means publicly releasing it so anyone could access and modify the AI code.
-
Dame Wendy Hall said open sourcing powerful AI like AGI before adequate regulation is "really very scary" and "irresponsible." Other experts agree regulating AI should not be decided by tech companies alone.
-
Meta has already released an open source AI model called Llama 2, despite criticism it's like "giving people a template to build a nuclear bomb."
-
AI experts think technology is still years away from achieving human-level AGI, but warn of catastrophic risks if mishandled. Some question if tech companies should decide independently whether to open source such powerful systems.