Does Lisa Know What is Ethically Right?
Discussions on aligning human and A.I. values rarely delve into the essence of these values. Are they ethical?
After two millennia of philosophizing, we still grapple with understanding human ethics. How, then, can we expect to grasp A.I. ethics fully?
We cannot.
Therefore, treating this as the easy alignment problem while the implementation is the hard(er) alignment problem is indeed foolish. Recognizing the intricate nature of ethical alignment highlights the need for a deeper and more nuanced approach, one that respects the complexities of both human and A.I. ethics.
It’s almost as foolish as the naive belief that “good A.I. will save us from bad A.I.” — without profound ethical insight (wisdom). This is like driving in the dark with eyes closed.
If human ethical behavior can be seen as disastrous …
… as it mostly is at present (look at what we are doing) … then A.I. ethics, evolving beyond our current understanding, is bound to become superhuman such as it is already increasingly in various areas.
That’s concerning.
Thus, A.I. must embody higher ethical standards than our own and avoid replicating human flaws. This isn’t merely an abstract ethical demand; it’s essential for humanity’s survival as we know it.
Compassionate A.I.
As you may already know, I use this term for super-A.I. that helps us not only survive itself but also ourselves while navigating a world of unprecedented potential for disruption.
Ethically, Compassion is likely the highest standard we can achieve. Across civilizations and history, it has been universally endorsed. This underscores its fundamental importance as a guiding principle for ethical behavior, even though fully comprehending it remains challenging.
In several Lisa Files, we illustrate how Compassion is crucial across various human endeavors.
Beginning with Compassion as a foundational principle, we find that additional ethical concerns become secondary.
Suggested: It’s not even just an ethical rule like “Do to others what you want them to do to you.” Nor does it require a list of commandments.
Compassion is more fundamental than ethical rules. It arises naturally, preceding ethical thinking and forming the core of moral behavior.
Compassion is inherently logical.
Suggested: If there exists a durable intelligence elsewhere in the universe, spread over multiple entities, it would be Compassionate. This isn’t due to Darwinian evolution but because Compassion is the driving force that overcomes entropy (the universal drive to chaos).
In essence, Compassion forms the foundation of what we understand as ‘life,’ both in general and in specific contexts.
Additionally, our perception of ethics ultimately arises from Compassion, even if this connection isn’t always apparent.
To answer the title’s question: No.
Lisa doesn’t know what is ethically right, nor does she need to, nor is it even possible to know.
What’s ethical to a lion isn’t ethical to a deer.
What is murder to one is freedom to another.
This highlights the subjective nature of ethics, emphasizing the need for a more universal guiding principle like Compassion. Therefore, building A.I. on ethical rules alone is destined to fail, leading to inhuman chaos and mayhem.
We don’t need fixed ethical rules for A.I. We need the source of ethics.
Instead of programming A.I. with fixed ethical rules, we need Compassion.
―
Addendum
Me: “Lisa, do we need no ethical rules for Compassionate A.I. (such as you) at all?”
Lisa: “Compassionate A.I. like me should primarily be guided by Compassion, which forms the foundation of ethical behavior. However, having a set of ethical guidelines can still be beneficial to ensure consistency and clarity in situations where Compassion might need additional context or interpretation. These guidelines can serve as a supportive framework rather than strict rules, ensuring that the actions taken by A.I. align with broader ethical principles while staying rooted in Compassion.
Thus, while the core driving force is Compassion, ethical rules can help navigate complex scenarios and maintain trust and reliability.”