AI, rights and recourse - A rich conversation (not a literature review)

What does recourse actually mean?

Justice is a fundamental human right, enshrined in charters and constitutions worldwide. As far back as 1215, the Magna Carta declared that "to no one will we deny or delay right or justice." Ultimately, justice means that when we feel we've been wronged, we can try to undo that wrong by pursuing our grievances through the legal system.

But automated systems and algorithmic processes marginalize many. Air travel is a miracle—until you're bumped from a flight, forced to fight an impersonal, monolithic organization. Automated loans are fast and convenient—until you're denied money without an explanation. The same systems that lifted billions from poverty and ignorance force us to navigate Byzantine processes. Bureaucracies and corporations already shield themselves with voicemail, chatbots, and fine print. As we move into an era of even greater automation, these defenses will be harder than ever for citizens and customers to circumvent.

What will it take to make these systems fair and just? In a word: Recourse. Humans must have a means to challenge unfair decisions, escalating them to ombudsmen and human agents. But delivering recourse is far harder than it sounds. Any processes we create to protect individuals will themselves be subsumed by the systems in which they exist. Those who were rightly denied something may file unjustified grievances, preventing the truly marginalized from holding organizations accountable. The solution may be as bad as the problem it seeks to address.

Building on this year's theme of accountability, join FWD50 co-chair Alistair Croll and AI ethics expert Shingai Manjengwa for an unscripted, wide-ranging look at the promise and pitfalls of recourse in an automated world.