Minds without conscience? Public perceptions of AI’s moral role

Among AI systems, Tesla’s Full Self-Driving car was rated highest in moral agency, comparable to a chimpanzee, while Roomba was perceived as having the least moral agency. This variation suggests that people judge AI based on their functional roles and appearances, leading to different levels of trust and responsibility attribution.

Artificial intelligence (AI) is no longer confined to science fiction. From chatbots and self-driving cars to advanced robotics, AI systems are increasingly integrated into society. However, how do humans perceive these entities in terms of intelligence, morality, and responsibility?

A recent study titled "Robots, Chatbots, Self-Driving Cars: Perceptions of Mind and Morality Across Artificial Intelligences", authored by Ali Ladak, Matti Wilks, Steve Loughnan, and Jacy Reese Anthis, published in the CHI Conference on Human Factors in Computing Systems, explores the public's perception of AI's mental and moral capabilities. Through a large-scale online study, the researchers uncover critical insights into how different AI systems are perceived and what implications this has for trust, accountability, and AI design.

Do AIs have minds? Understanding public perceptions

The study examines how 975 participants rated 14 AI systems and 12 non-AI entities on dimensions of mental agency (the ability to plan and act) and experience (the ability to sense and feel). The results indicate that people perceive AI to have low-to-moderate mental agency, ranking higher than inanimate objects but lower than animals like ants. AI systems were perceived to have particularly low experience, meaning people generally do not believe AIs can sense or feel emotions.

Interestingly, the study found significant variation in perceptions depending on the type of AI. Chatbots like ChatGPT and Replika were attributed more agency than robotic vacuum cleaners like Roomba, but still far less than a toddler or a chimpanzee. Among AI systems, Tesla's Full Self-Driving car was rated highest in moral agency, comparable to a chimpanzee, while Roomba was perceived as having the least moral agency. This variation suggests that people judge AI based on their functional roles and appearances, leading to different levels of trust and responsibility attribution.

The morality of machines: Assigning responsibility to AI

One of the most striking findings of the study is how people assign moral responsibility to AI. Moral agency (the capacity to do right or wrong) and moral patiency (the capacity to be treated rightly or wrongly) were measured across various AI entities. Participants attributed higher moral agency to AI than their mental experience would suggest. In simpler terms, while people do not see AI as "thinking" or "feeling," they do believe AI can be responsible for moral actions, especially in high-stakes contexts like autonomous vehicles.

Tesla's self-driving car, for example, was perceived to have as much moral agency as a chimpanzee, meaning participants saw it as capable of making ethical decisions on the road. On the other hand, AI with emotional or social components, like Replika or Sophia the humanoid robot, were perceived as having more moral patiency - deserving ethical consideration but not necessarily being responsible for moral actions. This discrepancy suggests that AI design influences public perception of responsibility, with potential consequences for AI ethics and governance.

Implications for AI design and policy

The findings of this study have far-reaching implications for AI design, governance, and regulation. If people attribute moral agency to AI systems, should AI developers and corporations be held accountable for AI-driven decisions? The study suggests that perceptions of AI can shape how responsibility is assigned in cases of AI-caused harm, such as accidents involving self-driving cars or biased hiring decisions made by AI algorithms.

Moreover, the research highlights the importance of transparency in AI design. If AI is perceived to have agency but little experience, it is crucial that users understand the limitations of these systems. Over-anthropomorphizing AI - making it appear too human-like - can lead to misplaced trust, while underestimating AI's decision-making capabilities might cause undue fear and rejection. Therefore, AI designers must carefully balance realism and transparency to create systems that are both effective and ethically aligned.

Future of AI morality: Navigating perception and reality

As AI systems become more integrated into daily life, understanding how people perceive their minds and moral responsibilities will be crucial for shaping AI policy and ethical guidelines. The study suggests that AI is perceived as something between an inanimate object and a living entity, creating a unique challenge in moral attribution.

Future research should explore how these perceptions change with AI advancements and how public attitudes toward AI morality influence legal and regulatory decisions. In the meantime, AI developers, ethicists, and policymakers must work together to ensure that AI is designed, deployed, and regulated in ways that align with both public perception and ethical responsibility. The challenge lies not just in how AI functions but in how we, as humans, understand and respond to its growing presence in society.