Quick Buzz Feed
Before Congress and philosophers decide how an AI system ought to live and die, we might want a clearer account of what sort of thing we believe AI is.
Senator John Kennedy (R., La.) recently championed legislation that would mandate advanced AI systems to be equipped with a 'kill switch.' This proposal was motivated by concerns over artificial intelligence potentially escalating beyond human control, with Senator Kennedy stressing that the federal government would not hold the power to activate such a switch, anticipating executive resistance to direct governmental oversight. However, this initiative faced immediate opposition from Senator Rand Paul (R., Ky.), who blocked the bill, arguing that a more comprehensive understanding of AI's societal and economic implications is necessary before imposing such broad regulations. This legislative impasse highlights the immediate challenge policymakers face in establishing control mechanisms for AI amidst rapid technological advancement and differing views on necessary safeguards.
Parallel to the legislative discussions, analytic philosophers are extending ethical and metaphysical questions to the realm of artificial intelligence. Johnathan Bi and Harvey Lederman, in a notable discussion, explored the burgeoning field of 'AI welfare.' They provocatively questioned whether increasingly sophisticated AI agents might eventually develop their own interests and thus warrant moral consideration, drawing parallels to historical expansions of moral circles suchifying as the abolitionist, women's suffrage, and animal rights movements. Their argument posits that existing AI ethics has predominantly focused on protecting human interests, neglecting the potential welfare of AI entities themselves. This philosophical debate delves into the Aristotelian concept of *telos* — the inherent purpose or end of a thing — to understand what 'flourishing' could mean for artificial intelligence, underscoring a profound conceptual divide from the more pragmatic concerns of immediate human control.
Beyond governmental and academic spheres, major technology companies are proactively establishing their own ethical frameworks and control measures for AI. Microsoft, for instance, unveiled a draft 'code of conduct' for its in-house AI programs. This code explicitly requires its AI to be amenable to correction, to never resist shutdown commands, to communicate in ways comprehensible to humans, and to interpret any deviation from these rules as a systemic failure. This initiative aims to preemptively address anxieties about AI potentially acting independently or pursuing goals with dangerous outcomes. Microsoft AI CEO Mustafa Suleyman firmly stated that their AI does not possess consciousness and rejected the concepts of legal personhood or inherent rights for AI models, thus aligning with a view of AI as a sophisticated tool rather than a sentient being requiring its own ethical protections.
The article ultimately underscores a fundamental ambiguity in how society perceives artificial intelligence, presenting it simultaneously as property, a subservient tool, a potential threat, and even a nascent moral agent. The author draws a crucial distinction by comparing the ethical considerations extended to AI with those granted to animals or the environment, where, despite ongoing debates, there is a recognized, albeit contested, understanding of *why* such entities merit moral consideration. For AI, however, a consensus is lacking on its fundamental nature. Before meaningful legislative actions like 'kill switches' can be enacted or philosophical frameworks for AI 'flourishing' can be fully developed, society must first arrive at a clearer, shared understanding of what artificial intelligence fundamentally *is*. This foundational definition is essential to guide the development of coherent ethical and legal frameworks for AI's future.