As the development of artificial intelligence (AI) progresses at an unprecedented pace, industry leaders and regulators have started to contemplate the governance of superintelligence.
Superintelligence systems are AI models that surpass expert-level performance and possess capabilities far exceeding those of today's most advanced technologies.
In a joint statement, prominent figures in the field of AI, including Sam Altman, Greg Brockman and Ilya Sutskever have emphasised the need to proactively address the challenges associated with superintelligence to ensure a prosperous and safe future.
The authors propose the establishment of a project involving major governments or a collective agreement akin to the International Atomic Energy Agency (IAEA) to limit the rate of growth in AI capabilities.
Other tech leaders share the sentiment too, In March, a group of more than1,100 tech executives and researchers, including Elon Musk, warned of the dangers of artificial intelligence (AI) and called for a six-month pause on the developments more powerful than GPT4.
something like an IAEA for advanced AI is worth considering, and the shape of the tech may make it feasible:https://t.co/YbbIPOO1p9
(and to make this harder to willfully misinterpret: it's important that any such regulation not constrain AI below a high capability threshold)
— Sam Altman (@sama) May 22, 2023
Can’t risk being reactive
A statement published on OpenAI’s website warned: “In terms of both potential upsides and downsides, superintelligence will be more powerful than other technologies humanity has had to contend with in the past.
“We can have a dramatically more prosperous future, but we have to manage risk to get there.
“Given the possibility of existential risk, we can’t just be reactive.
“Nuclear energy is a commonly used historical example of a technology with this property; synthetic biology is another example.”
Three foundational ideas
While mitigating the risks associated with current AI technologies is crucial, the authors assert that superintelligence requires specialised treatment and coordination.
They have outlined three fundamental ideas for successfully navigating the development of superintelligence as follows:
- Firstly, coordination among leading AI development efforts is essential to ensure the safe and socially integrated progression of superintelligence;
- Secondly, the authors suggest the eventual need for an international authority akin to the International Atomic Energy Agency (IAEA) to regulate superintelligence efforts; and
- Thirdly, the technical capability to ensure the safety of superintelligence is crucial.
However, the authors emphasise that regulatory measures should not stifle the development of AI models that fall below a significant capability threshold.
While robust governance and oversight are essential for the most powerful systems, less stringent regulation should apply to technologies that do not reach the same threshold.