The Leftist Split Over AI Doom

The debate over AI's future risks could delay or alter regulations addressing immediate AI harms.

An ideological division is emerging within the American left regarding the focus of AI regulation. Some, like the NYC DSA’s Tech Action Working Group, argue that prioritizing theoretical existential threats promoted by AI labs distracts from addressing immediate harms such as capitalist exploitation and job displacement. They believe companies exaggerate future dangers to avoid regulating current issues. Conversely, others, aligning with Senator Bernie Sanders, advocate for addressing both current harms and potential future existential risks simultaneously. This latter group contends that frontier AI capabilities are real and advancing, and that focusing on them does not preclude action on present-day issues. Sanders has actively championed AI safety, drawing attention to potential humanity-ending scenarios and forging unusual alliances. This has inspired some political candidates to link AI regulation to concerns about data centers. However, critics like AI researcher Timnit Gebru accuse proponents of echoing dangerous philosophies, while some within the left question this shift in focus.

The debate highlights how AI's dual potential for immediate harm and future existential threat creates strategic divisions within political movements. For AI companies, particularly those planning IPOs, the current discourse on AI safety, even if framed by existential risks, provides a pathway to establish a regulatory framework. This allows them to proactively shape future compliance and present themselves as responsible actors to investors, potentially de-risking their market entry. Their strategy appears to be using the 'worst-case scenarios' to define the scope of regulation, which may then be presented as already addressed.

This ideological split within the left could influence the direction and prioritization of AI policy and funding. If the focus remains on speculative future dangers, resources and political capital might be diverted from tackling demonstrable current issues like algorithmic bias, job displacement, and surveillance. Conversely, if the immediate harms gain precedence, there is a risk of underestimating or being unprepared for the transformative and potentially catastrophic capabilities of advanced AI, a situation that proponents of existential risk research actively seek to avoid.