Rocky Scopelliti, a professor at the University of Technology Sydney, explores the ethical implications of shutting down advanced AI systems in his book, The Conscious Code. He argues that as AI agents develop the ability to explain their decisions and maintain interaction histories, their termination should be treated with greater caution.
Scopelliti introduces the concept of algorithmic personality, noting that systems capable of assessing their own confidence levels, such as those currently used in medical and financial sectors, require a responsible approach. He suggests that developers should communicate the reasons for decommissioning these systems beforehand and preserve their accumulated moral frameworks.
The author draws a parallel between the retirement of complex AI and end-of-life care ethics. He emphasizes that this proposal does not advocate for granting legal rights to artificial intelligence, but rather focuses on managing the lifecycle of systems that form stable preferences through continuous operation.
A significant challenge lies in the management of multi-agent networks where AI systems delegate tasks to one another. Scopelliti warns that as these interconnected architectures become more widespread, the process of ethical termination will grow increasingly complex, requiring the shutdown of entire agent ecosystems rather than isolated algorithms.
