Aubrey de Grey discusses the goals of artificial intelligence (AI) research, emphasizing the importance of building AI systems with common sense rather than pursuing superintelligent, recursively self-improving machines. Here are the key points:
1. Common Sense AI: De Grey believes that the primary goal of AI should be to create machines with common sense, often referred to as “minions.” These AI systems would possess enough common sense to perform tasks with minimal supervision.
2. Concerns about Superintelligent AI: Some AI researchers aim to build machines with capabilities vastly exceeding human cognition. De Grey acknowledges the potential benefits but also highlights concerns about the risks of recursive self-improvement, where AI systems improve themselves in ways beyond human understanding and control.
3. The Unlikelihood of Recursive Self-Improvement: De Grey suggests that recursive self-improvement may be mathematically impossible. He argues that there might be a limit to how much a program can improve itself, and the program it writes may be simpler than itself. This would inherently limit the process of recursive self-improvement.
4. Ethical Concerns: De Grey expresses ethical concerns about creating creative machines, whether technologically or artistically, as it blurs the distinction between humans and machines. He is particularly worried about determining the rights and moral judgments for entities that differ significantly from humans.
5. Common Sense Definition: De Grey defines common sense as the ability to process incomplete information to identify an optimal method for achieving a specified goal from a predefined set of methods. Common sense excludes the option of seeking new methods outside the predefined set.
6. Avoiding Recursive Self-Improvement: De Grey believes that it’s possible to prevent recursive self-improvement by designing AI systems that adhere to predefined sets of methods and goals. This would eliminate the risk of “method creep.”
7. Ethical Considerations: De Grey calls for proactive consideration of the rights of thinking machines. He suggests that society should explore the ethical implications of AI systems with limited awareness of their goals before developing more advanced AI with recursive self-improvement capabilities.
In summary, Aubrey de Grey advocates for the development of AI systems with common sense and highlights concerns about superintelligent AI. He calls for ethical discussions on the rights of thinking machines to address potential challenges and consequences in the future.