David Manheim

This paper examines some limitations of large language models (LLMs) through the framework of Peircean semiotics. We argue that basic LLMs exist within a “hall of mirrors,” reflecting only the linguistic surface of training data without indexical grounding in a shared external world, and manipulating symbols without participation in socially-mediated epistemology. We then argue that newer developments, including extended context windows, persistent memory, and mediated interactions with reality, are moving towards making newer Artificial Intelligence (AI) systems into genuine Peircean interpretants, and conclude that LLMs may be approaching this goal, and we identify no fundamental architectural barriers that would prevent this. This lens reframes a central challenge for AI alignment: without grounding in the semiotic process, a model’s linguistic encoding of goals may diverge from real-world values. By synthesizing Peirce’s pragmatic view of signs, contemporary discussions of AI alignment, and recent work on relational realism, we illustrate a fundamental epistemological and practical challenge to AI safety and point to part of a solution.

This paper examines some limitations of large language models (LLMs) through the framework of Peircean semiotics. We argue that basic LLMs exist within a "hall of mirrors," manipulating symbols without indexical grounding or participation in socially-mediated epistemology. We then argue that newer developments, including extended context windows, persistent memory, and mediated interactions with reality, are moving towards making newer Artificial Intelligence (AI) systems into genuine Peircean interpretants, and conclude that LLMs may be approaching this goal, and no fundamental barriers exist. This lens reframes a central challenge for AI alignment: without grounding in the semiotic process, a models’ linguistic encoding of goals may diverge from real-world values. By synthesizing Peirce's pragmatic view of signs, contemporary discussions of AI alignment, and recent work on relational realism, we illustrate a fundamental epistemological and practical challenge to AI safety and point to part of a solution.

Auditing of AI systems is a promising way to understand and manage ethical problems and societal risks associated with contemporary AI systems, as well as some anticipated future risks. Efforts to develop standards for auditing artificial intelligence (AI) systems have therefore understandably gained momentum. However, current approaches are not just insufficient, but can be actively harmful. Transparency alone does not address concerns about risk. Internal auditing is insufficient, and easily becomes safety-washing. External audit is better, but requires credible standards. Industry-led approaches to building standards or to perform audits lack credibility and undermine other efforts. Regulation often is ill adapted and becomes a static barrier. Lastly, all of these limited technical, governance, and even ethical assessments fail to ensure continued stakeholder input and engagement. Instead, the paper proposes the establishment of an AI Audit Standards Board, in line with best practices in other fields, including safety-critical industries like aviation and nuclear energy, as well as more prosaic ones such as financial accounting and pharmaceuticals. This would address the evolving nature of AI technologies, help maintain public trust in AI, and promote a culture of safety and ethical responsibility within the AI industry. By ensuring audits remain relevant, robust, and responsive to the rapid advancements in AI, auditing AI will not devolve into safety washing and addresses risks and ethical concerns that will continue to arise as AI becomes increasingly important in society, and as human interaction with these systems changes over time.

This paper considers challenges resulting from the use of advanced artificial judicial intelligence (AAJI). We argue that these challenges should be considered through the lens of value alignment. Instead of discussing why specific goals and values, such as fairness and nondiscrimination, ought to be implemented, we consider the question of how AAJI can be aligned with goals and values more generally, in order to be reliably integrated into legal and judicial systems. This value alignment framing draws on AI safety and alignment literature to introduce two otherwise neglected considerations for AAJI safety: specification and assurance. We outline diverse research directions and suggest the adoption of assurance and specification mechanisms as the use of AI in the judiciary progresses. While we focus on specification and assurance to illustrate the value of the AI safety and alignment literature, we encourage researchers in law and philosophy to consider what other lessons may be drawn.

How much value can our decisions create? We argue that unless our current understanding of physics is wrong in fairly fundamental ways, there exists an upper limit of value relevant to our decisions. First, due to the speed of light and the definition and conception of economic growth, the limit to economic growth is a restrictive one. Additionally, a related far larger but still finite limit exists for value in a much broader sense due to the physics of information and the ability of physical beings to place value on outcomes. We discuss how this argument can handle lexicographic preferences, probabilities, and the implications for infinite ethics and ethical uncertainty.

The possibility of social and technological collapse has been the focus of science fiction tropes for decades, but more recent focus has been on specific sources of existential and global catastrophic risk. Because these scenarios are simple to understand and envision, they receive more attention than risks due to complex interplay of failures, or risks that cannot be clearly specified. In this paper, we discuss the possibility that complexity of a certain type leads to fragility which can function as a source of catastrophic or even existential risk. The paper first reviews a hypothesis by Bostrom about inevitable technological risks, named the vulnerable world hypothesis. This paper next hypothesizes that fragility may not only be a possible risk, but could be inevitable,and would therefore be a subclass or example of Bostrom’s vulnerable worlds. After introducing the titular fragile world hypothesis, the paper details the conditions under which it would be correct, and presents arguments for why the conditions may in fact may apply. Finally, the assumptions and potential mitigations of the new hypothesis are contrasted with those Bostrom suggests.