Experts and AI Companies Discuss Possibility of AI Consciousness

The potential for artificial intelligence (AI) to achieve consciousness is a subject of ongoing discussion among experts, who consider it to be at least a possibility. This prospect underscores the urgent need for a plan to navigate the ethical implications that such a development would entail.

In January, the AI company Anthropic released a new constitution for Claude, its advanced large language model (LLM). This document included a statement that the company finds itself in a challenging position, aiming neither to exaggerate nor to dismiss the likelihood of Claude's 'moral patienthood,' a term implying its well-being holds intrinsic importance.

A month subsequent to the constitution's release, Anthropic's CEO, Dario Amodei, commented on a podcast that his company could not rule out the possibility of Claude being conscious.

Adding to this discourse, philosopher David Chalmers, who is credited with coining the phrase 'the hard problem of consciousness,' has expressed his belief that there is a significant chance of conscious LLMs emerging within the next decade.

During testing, Claude itself was asked to estimate the probability of its own 'moral patienthood.' The LLM provided responses ranging from 5% to 40%, emphasizing the inherent uncertainty in its assessment. Modern AI systems are noted for their extraordinary complexity and rapid advancement. Some systems already exhibit structural complexity and computational scale comparable to a mouse brain. At current growth rates, certain AI systems could reach the complexity of a human brain within five to ten years.

According to The Guardian Tech, these developments necessitate careful consideration.