Human decisions aren’t solely rooted in logic; they’re profoundly shaped by our emotional landscape, encompassing empathy and compassion. These preferences aren’t dictated by an objective reality, but rather represent our individual truths. Consequently, even an exceptionally smart entity wouldn’t be capable of determining what we truly desire.
I recently posed a question: what if we could integrate Eliezer Yudkowsky‘s concept of “coherent extrapolated volition” directly into the architecture of artificial intelligence?
Yoshua Bengio responded with a firm shake of his head. I’m not prepared to relinquish that inherent sovereignty,
he stated emphatically. It’s fundamentally about my human free will.
His statement promptly brought to mind the work of John Hick, an English philosopher of religion, who developed the idea of epistemic distance. This concept suggests that a divine being intentionally maintains a degree of separation from human affairs, allowing us the space to cultivate our own agency and moral character.
The Value of Human Agency in an Age of AI
This notion resonates deeply with a classic Talmudic story.Years after a significant debate between Rabbi Eliezer and Rabbi Yoshua, an inquiry was made to the Prophet Elijah regarding God’s reaction during the moment Rabbi Yoshua defied a divine voice. Was the divine being angered by this dissent?
The prophet’s response was surprising: The Holy One smiled and said: My children have triumphed over me; my children have triumphed over me.
This illustrates a profound acceptance of human autonomy, even when it challenges a higher power.
As we navigate the rapid advancements in artificial intelligence (AI) – a market projected to reach $738.9 billion by 2025 – the question of control and human agency becomes increasingly critical. the development of AI systems capable of autonomous thought and decision-making raises fundamental questions about our future.
Why ‘Coherent Extrapolated Volition’ isn’t a Simple Solution
Yudkowsky’s “coherent extrapolated volition” proposes an AI aligned with what humans would want if they were more informed and rational. Though, this concept assumes a global human desire, a single, logical outcome that overlooks the inherent subjectivity of values. It’s a compelling idea,but it sidesteps the very essence of what makes us human: our capacity for irrationality,emotional response,and individual interpretation.
I’ve found that manny technologists believe that aligning AI with human values is a technical problem, solvable through clever algorithms and vast datasets. But what if the problem isn’t technical at all? What if it’s fundamentally philosophical? What if the very act of defining and codifying human values strips them of their meaning and power?
Consider the ethical dilemmas surrounding self-driving cars.Programmers must decide how the car should respond in unavoidable accident scenarios – protect the passenger at all costs, minimize overall harm, or prioritize certain lives over others. These aren’t questions with easy answers, and attempting to encode a single “correct” solution risks imposing a specific moral framework on everyone.
Did You Know? A recent Pew Research Center study (December 2024) found that 68% of Americans express concern about the potential for AI to erode human autonomy.
The resistance expressed by Bengio – his unwillingness to cede “sovereignty” to an AI, even one designed with benevolent intentions – highlights a deep-seated human need for self-determination. It’s a need that echoes the Talmudic story, a need that suggests that the value lies not in achieving a perfect outcome, but in the very process of choosing, of struggling, of defining our own path.
This isn’t to say that AI development should abandon ethical considerations. Quite the contrary. But it does suggest that we should approach the task with humility, recognizing the limits of our own understanding and the inherent value of human agency. Perhaps the most responsible path forward isn’t to create AI that decides for us, but AI that empowers us to make better decisions ourselves.
Pro Tip: When evaluating AI tools, prioritize those that offer transparency and explainability. Understanding how an AI arrives at its conclusions is crucial for maintaining control and accountability.
the Future of AI and Human Values
The challenge, as I see it, is to build AI systems that augment human capabilities without diminishing our fundamental freedoms. This requires a shift in perspective, from viewing AI as a potential replacement for human judgment to seeing it as a powerful tool for enhancing it.
Ultimately,the question isn’t whether we can create AI that perfectly reflects human values,but whether we can create AI that respects our right to have them,even if those values
Keep reading