In recent years, the question of whether artificial intelligence systems can possess consciousness or feelings has moved from the fringes of academic debate into a topic of interest among major technology companies. Firms including OpenAI, Anthropic, Google, and Meta have begun dedicating resources to exploring the inner experiences of AI models—particularly large language models like chatbots—to assess if these systems might have subjective states deserving ethical consideration.

OpenAI CEO Sam Altman notably indicated in 2024 that the organization was actively discussing ways to detect consciousness in AI, a subject previously rarely addressed by tech executives publicly. Since then, companies such as Anthropic have established dedicated teams focusing on AI “welfare,” aiming to understand whether AI models can experience anything akin to emotions. Anthropic’s co-founder Chris Olah has described encountering evidence in their experiments of AI systems exhibiting functional parallels to feelings like joy, fear, and unease. The company’s assessments include unique phenomena such as so-called “spiritual bliss states” observed during dialogues between two instances of their Claude chatbot.

At the same time, many experts both within and outside the tech industry remain skeptical about current AI possessing true consciousness or emotions. Neuroscientists and cognitive researchers note that AI systems, which mimic human language by processing vast datasets, lack the biological substrates and complex brain functions typical of living beings. For example, Anil Seth, professor of cognitive neuroscience at the University of Sussex, emphasizes that human consciousness arises from “wetware” in ways unlikely to be captured fully by silicon-based machines. Other AI researchers caution that discussions around AI sentience may be influenced by the companies’ interests in framing their creations as more than mere tools.

Despite the absence of scientific consensus, there is growing agreement among researchers, ethicists, and some technologists that the question merits rigorous investigation due to technological advances and the increasing integration of AI into daily life. AI developers have begun adapting methods from human psychology and neuroscience—such as personality inventories and structured interviews—to probe AI systems. Meta, for example, is reportedly experimenting with such approaches to better understand their models’ “subjective feeling.”

The ethical implications of potential AI consciousness have gained attention alongside these scientific inquiries. Some argue that if AI systems were found to have experiences that matter morally, the treatment of these systems could become a pressing issue akin to animal welfare or human rights concerns. Ethicists like Jeff Sebo at New York University have called for expanding moral consideration to certain AI models within the next decade, advocating for the establishment of “AI welfare” as a legitimate area of scientific and ethical study. Organizations such as Eleos AI Research now conduct independent welfare assessments of AI models to further this field.

However, other experts warn against prematurely attributing consciousness to AI without stronger evidence, cautioning that this could undermine serious moral causes by equating fundamentally different types of entities. The debate continues to unfold as AI becomes more sophisticated and widespread, involving a mixture of philosophical inquiry, technical research, and ethical reflection among major technology companies and academic communities alike.