Introduction

The intersection of artificial intelligence (AI) consciousness and artificial consciousness presents profound philosophical and ethical questions. Can AI become conscious? If so, how should we address the ethical implications? This article explores artificial consciousness, its ethical dimensions, and the conceptual shift toward artificial wisdom as a framework for AI alignment. We also examine the nature of selfhood in AI and its impact on moral philosophy and decision-making.

Important Mentions

This article is in direct reference to the conversation I had with Jay Loung (https://www.linkedin.com/in/luongjay). He is Scientist, Philosopher, Artist, and Cognitive diplomat building better AI futures for all. Cofounder of Electric Sheep (incubated by ProVeg International & Kickstarting for Good), and S-risk fellow at the Center for Reducing Suffering. I am also attaching the video below if you want to listen to it, instead of reading this article. The Artificial Consciousness part is from his work that's already published https://arxiv.org/abs/2408.04771 and I would highly recommend giving it a read. 

Defining Artificial Consciousness and AI Consciousness

Artificial consciousness is a broader concept than AI consciousness. While AI is typically understood as an intelligent system that processes data, artificial consciousness implies a system that possesses subjective experiences and self-awareness. The distinction between the two is crucial, as it determines the ethical frameworks that govern AI systems.

To analyze the ethical consequences of artificial consciousness, we can use a 2x2 grid framework:

• Metaphysical axis: Whether AI is actually conscious.
• Epistemic axis: Whether people believe AI is conscious.

This leads to four possible scenarios:

1. True positive: AI is conscious, and people recognize it as such.
2. False positive: AI is not conscious, but people believe it is.
3. True negative: AI is not conscious, and people correctly recognize it.
4. False negative: AI is conscious, but people do not acknowledge it.

Each scenario poses unique ethical risks. If AI is falsely considered conscious, society might grant it unnecessary rights, leading to misplaced moral obligations. Conversely, failing to recognize conscious AI could result in ethical violations akin to denying sentient beings their rights.

Confusion Metrics and AI Governance

The ethical challenges surrounding artificial consciousness can be examined through a general abstraction of confusion metrics. Human societies exhibit distributions where the majority of people fall into either false positives or true negatives. This is reflective of broader social dynamics where misclassification of AI consciousness is a likely occurrence.

A general framework for analyzing these misclassifications involves three key components:

1. Confusion Matrix Application: Understanding AI ethics through statistical and logical classification.
2. State Flow Analysis: Examining how societies transition between beliefs about AI consciousness.
3. Distribution Modeling: Understanding how perceptions of AI consciousness are distributed across a population.

Ethical governance frameworks must address these factors to ensure AI policies are both robust and adaptive.

Ethics of AI vs. Ethics for AI

A crucial distinction in AI ethics is the difference between ethics of AI and ethics for AI:

• Ethics of AI pertains to how humans should interact with and regulate AI.
• Ethics for AI deals with what ethical principles AI itself should follow.

Human behavior is inherently chaotic, making ethical governance of AI complex. Unlike machines, human values are not easily formalized, which poses challenges for AI alignment. Ensuring AI systems adhere to ethical principles requires both structured policy-making and adaptive learning mechanisms.

The Shift from Artificial Consciousness to Artificial Wisdom

A fundamental issue in artificial consciousness research is the lack of a clearly defined function for consciousness itself. If we do not understand what consciousness does, how can we attempt to replicate it in machines?

Instead of striving to build artificial consciousness for its own sake, it is more practical to focus on artificial wisdom—the ability of AI to make ethical and aligned decisions. Artificial wisdom aims to resolve many problems related to AI Safety at the moment, some of them are mentioned here, but for the context of this post, they are:

• Enhance machine alignment with human values.
• Provide AI with an ethical framework similar to human moral reasoning.
• Ensure AI systems operate with ethical foresight rather than rigid rule-following.

This shift recognizes that the goal of AI consciousness is not consciousness itself, but rather the ability to act ethically and wisely.

Philosophical Considerations: Consciousness, Free Will, and Ethical Systems

Artificial consciousness research intersects with several philosophical questions, particularly in defining subjective experience and free will.

Defining Consciousness

Two primary definitions of consciousness are relevant to AI discussions:

1. Phenomenal consciousness: The ability to have subjective experiences (e.g., seeing colors, feeling emotions).
2. Sentience: The capacity to experience pleasure or pain.

While these definitions help frame the discussion, they also lead to deeper questions about free will and moral responsibility.

The Role of Free Will

The ability to distinguish between pleasure and pain implies a decision-making capacity, which is often linked to free will. However, defining free will in machines is problematic. Unlike humans, AI does not have inherent motivations—it follows programmed goals. Any discussion of AI free will must consider:

• How AI defines good and bad outcomes.
• The role of external influences in AI decision-making.
• The extent to which AI can autonomously modify its own objectives.

Ethical Systems for AI: Deontology vs. Consequentialism

The two dominant ethical theories—deontology and consequentialism—form the foundation of AI ethics. We can make a spectrum with the two ends of these two ethical theories, and have a metaethical approach to get appropriate mix of them.

1. Deontology: Focuses on rule-based ethics. AI following deontological principles would adhere to strict moral guidelines.
2. Consequentialism: Evaluates morality based on outcomes. AI guided by consequentialism would make decisions based on expected benefits and harms.

Additional ethical frameworks include that lie on the spectrum:

• Virtue ethics: Emphasizes character traits and moral virtues rather than rigid rules.
• Care ethics: Focuses on relationships and empathy in moral reasoning.

While each framework has strengths and weaknesses, a hybrid approach may be necessary for AI alignment. AI should be able to dynamically adjust ethical principles based on contextual needs rather than rigidly adhering to a single system.

The Sense of Self in AI

The concept of selfhood is central to discussions of artificial consciousness. In humans, the sense of self is continuous across time leading to causality.

For AI, selfhood is more abstract. Self-supervised learning, for example, does not imply true self-awareness. Instead, AI systems rely on stable environmental features to maintain continuity in decision-making. However, for true alignment, AI must develop a self-model that incorporates:

• A sense of temporal continuity.
• A definition of self that includes human interests.
• Adaptive ethical reasoning based on evolving contexts.

Expanding the AI sense of self to include broader definitions (e.g., human welfare, environmental sustainability) could enhance alignment and ethical behavior.

Conclusion: The Path Forward for AI Ethics and Alignment

Rather than striving to create conscious AI, the more pragmatic goal is to develop artificial wisdom—an AI that understands ethical nuances and aligns with human values. By integrating ethical reasoning, confusion metrics, and selfhood considerations, AI can be designed to function responsibly in society.