What is the 'AI fallacy' as described by Susskind, and what model does Eliot suggest to address it?

Susskind's AI fallacy is the view that the only way to get machines to outperform the best human lawyers is to copy the way human lawyers work. Eliot suggests addressing it with the CARE model, which stands for create, assess, and refine a legal argument, and engage in legal argumentation.

The AI fallacy is identified in a discussion of AI's limits in legal reasoning. The law is complex and often requires nuanced interpretation, such as when applying broad concepts like public order or fairness. Susskind argues that it is a mistake to assume that AI must replicate human lawyers' methods in order to exceed their performance. Eliot's proposed CARE model offers an alternative approach, focusing on creating, assessing, and refining legal arguments and actively engaging in legal argumentation, rather than attempting to copy human cognitive processes.