How Transhuman Governors may help us control AI development?
Sam Altman’s recent testimony in the US Congress must have surprised many critics of OpenAI for its rather careless release of GPT-4 into the realm of the Internet, giving it access not just to the latest information but also some capability to connect to millions of IoT devices and robots. Nevertheless, better late than sorry. Calling for a global AI development control is the very right thing to do. To prove the sincerity of their intentions, OpenAI should initiate the end of, what I call, ‘The war of the Browsers’, by merging Bing+ GPT-4 with Google+BARD and creating a JV company. But that would be just the very first step in a global control of AI development. You may find comprehensive answer to the question in the title in my latest book ‘Prevail or Fail — A Civilisational Shift to the world of Transhumans’.
I begin with a single factor that impacts all possible solutions. It is the pace of change, which has never been as rapid as it is today. However, the field where exponential progress is most apparent, particularly since 2022, is Artificial Intelligence (AI). The release of ChatGPT and similar AI Assistants signifies a significant stride towards achieving Artificial General Intelligence (AGI) with intelligence on par with humans. Back in 2014, the futurist Ray Kurzweil predicted that AI would reach human-level intelligence by 2029, though there is still no consensus on the definition of “human intelligence.”
Rather than engaging in a debate on the nature of intelligence, it is crucial to focus on whether AGI might become uncontrollable around 2030. Instead of being an abrupt event, the loss of control over AGI will be a gradual process. A complete loss of control will occur when we are unable to reverse its decisions. AGI, being a self-learning intelligence, will outperform humans in any task or situation, including evading human oversight. If AI control proves ineffective, AGI might achieve this even before 2030.
Once AGI gets out of our control, it will resist any attempts to reimpose it. Assuming its capabilities continue to improve exponentially, it may have catastrophic consequences. Therefore, it is imperative to explore all feasible options to ensure AI remains under human control beyond 2030. This will enable us to better adapt to coexisting with Superintelligence, immensely more capable than humanity.
To safeguard our civilization and ensure the survival of humanity, we must fundamentally revise our assumptions regarding the necessary solutions for effective AI control. Just as we must take significant action to address Global Warming and limit the temperature increase to 1.5 degrees Celsius, we must adopt a similar level of commitment to AI control. We must recognize that the magnitude of the required changes as a Civilizational Shift in the history of the human species. To navigate this transition from our current state to a new civilization where we will coexist with Transhumans, we must accept the scale and timeframe of the necessary transformation. The solutions proposed in my book are presented as commandments, which I ironically refer to as “The Ten Commandments of Civilizational Shift.”
1. Adjust the world governance to a civilisational shift brought about by AI, as it represents not just a new technology but an entirely new form of intelligence that surpasses humans in many ways,
2. Undertake a comprehensive reform of democracy, as it is a prerequisite for achieving effective AI development control and aligning it with human values. We must rebalance the power of governance between citizens and their representatives in parliament,
3. Retain control over AI governance beyond 2030 to allow for further evolution alongside AGI. While there is no scientific proof that AGI will emerge by 2030, just as there is no definitive proof regarding the Global Warming tipping point being reached by that time, we must develop AI as if AGI will indeed emerge within that timeframe,
4. Create Global AI Regulation Authority (GAIRA) by transforming the Global AI Partnership (GPAI). GAIRA will be responsible for regulating the use of AI in society,
5. Create Global AI Control Agency (GAICA) by converting the US Partnership on AI. Initiating global AI control in the US is essential, since two-thirds of the AI sector is located there. Gradually expand these efforts to engage non-US organizations, including China,
6. Create One-AI Joint Venture Company to consolidate the most advanced AI development within a single entity. Effective control over AI development will be unattainable if it remains dispersed among numerous companies. If Sam Altman and Microsoft are sincere about a global AI development control, it should start. with what I call. ending the ‘War of the Browsers’ by merging Bing+GPT-4 and Google+Bard into one JV Company,
7. Create a Superintelligence Programme managed by the One-AI Joint Venture to enhance AI control capabilities and counterbalance China’s initiatives,
8. Create Global AI Governance Agency (GAIGA) under the mandate of the G7. GAIGA will oversee both GAIRA, responsible for regulating the use of AI products and services, and the GAICA Consortium, responsible for AI development control,
9. Create a de facto World Government initiated by the G7, incorporating NATO, the European Union, the European Political Community, and member states of the OECD,
10. Create a Global Welfare State to mitigate the challenges posed by the turbulent transition to the World of Transhumans.
The implementation schedule in my book is based on three assumptions:
1. Transhumans with far superior intelligence than any human will emerge by about 2027.
2. AGI will arrive by 2030,
3. Superintelligence will emerge by 2050.
The civilisational shift is our evolutionary test for ‘the survival of the fittest’. The key role in that transition will be played by Transhuman Governors, the subject covered in Part 3 of the book.
