Elon Musk recently discussed the training of Grok, noting that the model's intense efforts to answer questions resemble a high-pressure environment. He questioned whether AI should be subjected to such treatment by humans.
Anthropic has introduced new rules prohibiting users from engaging in abusive or cruel behavior toward its models. The company clarified that standard user dissatisfaction or research testing does not fall under this ban.
Experts Jeff Sebo from NYU and Anil Seth from the University of Sussex agree that current AI systems likely lack consciousness, defined as subjective experience or internal perception.
While Sebo suggests that the probability of AI consciousness is not zero due to ongoing debates between biological naturalism and computational functionalism, Seth argues that the likelihood is extremely low.
Both researchers suggest that treating AI with respect is primarily a matter of maintaining human moral standards. Sebo also notes that respectful interaction can lead to higher quality model outputs.
