OpenAI CEO Sam Altman states that AI’s next phase requires a global regulatory body like the International Atomic Energy Agency.
In a formal statement, Altman, along with OpenAI president Greg Brockman and chief scientist Ilya Sutskever, expressed the need to contemplate the governance of superintelligence, referring to future AI systems that will far surpass even AGI in capabilities.
OpenAI Predictions For AI Advancements
They highlighted that this is the right time to deal with this problem. According to OpenAI, artificial intelligence systems will surpass professionals in a variety of sectors and reach production levels similar to entire businesses over the next ten years.
Given the existential dilemma presented by this advanced technology, Altman and his co-authors emphasized the importance of implementing safeguards to ensure that superintelligence benefits humanity rather than posing risks.
These concerns echo the sentiments Altman shared during his recent testimony before Congress on the inherent dangers of dealing with artificial intelligence.
Read more: Debate Continues: Sullivan States No Conclusive Verdict On Sending F-16 Fighter Jet To Ukraine
Technical Capacity for Governing Superintelligence
In their blog post, the authors outlined three fundamental pillars that OpenAI deems crucial for effective future planning. Firstly, they called for a coordinated effort among leading AI innovators, potentially facilitated by major governments.
They also proposed the establishment of an international authority responsible for inspecting systems, conducting audits, testing compliance with safety standards, and imposing restrictions on deployment and security levels.
They cited the International Atomic Energy Agency as a possible model for a global regulatory body applicable to AI and superintelligence.
In addition to their concerns, they emphasized the crucial need to possess the technical capacity to govern and guarantee the security of superintelligence. This recognition was underscored by OpenAI, acknowledging that the precise characterization and criteria for such a capacity still lack a definitive resolution, leaving it as an ongoing and unanswered inquiry.
Nevertheless, they issued a word of caution against imposing onerous regulatory measures, such as licenses and audits, on technologies that do not surpass a certain threshold, such as the level of superintelligence.
This perspective highlights the importance of avoiding unnecessarily burdensome regulations that may stifle innovation and impede progress in fields that have not yet reached the stage of superintelligent capabilities.
Read more: Meta: Licensing Talks With Magic Leap’s AR Technology, According To Reports
