Personhood is the best solution to responsibility problems.

Harsh Amrit and Nitya Malviya 21, “Legal Personhood to AI and Human Rights: An Analysis,” The Human Rights Blog, 05-26-2021, https://casihrrgnul.wordpress.com/2021/05/26/legal-personhood-to-ai-and-human-rights-an-analysis/

AI can be most inclusively defined as, a system that can think and act, based on its experience[i]. AIs are getting technologically advanced and gaining the capability to act independently (Tesla self-driving car, robotic nurses, virtual assistants, etc.) Till now, such systems have been treated as objects of law and any act of the system leading to damage would make the producers and the programmer liable. However, with the advent of newer and more sophisticated AI, determining liability has become a challenge. There is an increasing need to treat AI as subjects of law by granting them legal personhood. As it stands, however, people and institutions are still accountable for the actions of AI. As per the UNCITRAL, the person on whose behalf the Computer was working would be liable in the case of an unwanted circumstance. But, in the case of an advanced system, where multiple AI is doing the job, it would be difficult to trace the person responsible, leading to a responsibility gap.[ii] Like in high-frequency trading, where a large number of financial transactions are performed in fractions of a second by AI, the situation would be unusual for determining the responsibility if something illegal happens. Therefore, granting personhood to AIs would ensure smoothness of work and address the responsibility gap.