Human Goals Are Constitutive of Agency in Artificial Intelligence (AI)

“While I do not disagree with Matthias (2004) in holding that people should not be blamed for things they had no full control of (in this case, involving operations performed by machines), I argue for a more complex picture, taking social and normative factors into consideration that went into…”

“In criticizing this view, Johnson (2006) points out that even though moral agents in general may exhibit these features, one cannot ignore that artificial systems were designed by humans and perform their functions within a social setting. As such, choosing this specific level of abstraction appears to miss one centra…”

“The central normative claim in this context is that AI should align with human goals, and this point has been made in ethical context since the beginnings of AI research (Wiener, 1960), with a more recent formulation in Asilomar and Principles, 2017).”

“These agents are defined by their ability to perceive structured or unstructured inputs [46], to reason over contextual information [47,48], and to initiate actions toward achieving specific objectives, often acting as surrogates for human users or subsystems [49].”