Much of the time, conversations regarding artificial intelligence often veer away from the practical applications in today’s world. Earlier this year, leaders at Anthropic, Google DeepMind, OpenAI, and other AI firms emphasized in a collective statement that addressing the risk of extinction from AI should be a global priority on par with other large-scale societal risks like pandemics and nuclear conflict. British Prime Minister Rishi Sunak, ahead of the recent AI summit he hosted, cautioned about the potential loss of control over AI. Existential risks, also known as x-risks in AI circles, evoke images from blockbuster science-fiction movies and tap into deep-seated fears.
However, AI already presents economic and physical dangers, particularly impacting the most marginalized members of society. Some individuals have faced incorrect denial of healthcare coverage or unjust detention based on algorithms claiming to forecast criminal behavior. Certain applications of artificial intelligence, such as AI-driven target-selection systems used by the Israeli military in Gaza, directly endanger human lives. In other instances, governments and corporations leverage AI to disempower the public and mask their true intentions through methods like embedding austerity measures in unemployment systems, implementing worker-surveillance technologies to diminish autonomy, and utilizing emotion-recognition systems with flawed foundations to influence hiring decisions.
During Sunak’s summit, the AI Now Institute, along with a few other watchdog organizations, participated in discussions where global leaders and tech executives deliberated on hypothetical threats to “humans” devoid of race or gender in a distant future. This setting highlighted the insular nature of most AI deliberations.
The concept of artificial intelligence has evolved over the past seven decades, with the current iteration heavily influenced by the substantial economic dominance of major tech corporations in recent years. The resources necessary for large-scale AI development—extensive datasets, computational power, and skilled labor—are predominantly concentrated among a select few firms. Moreover, the field’s direction is primarily steered by the profit-driven motives of industry players rather than the broader public interest.
A headline in the Wall Street Journal this summer, “In Battle With Microsoft, Google Bets on Medical AI Program to Crack Healthcare Industry,” sheds light on the competition between tech giants to create chatbots aiding healthcare professionals, especially those in underserved medical settings, in accessing information swiftly. Despite the industry’s prowess in launching products that cater to the majority, there remains a significant failure rate for marginalized groups. This issue is particularly critical in healthcare applications, where safety standards must be rigorously upheld.
The narrative around AI is often dominated by hypothetical dangers posed by “frontier AI,” deflecting attention from the immediate risks faced today. The Biden administration, in contrast to Sunak, is placing a stronger focus on addressing present threats. The recent executive order from the White House encompasses provisions addressing various aspects of AI’s impact on society, including competition, labor rights, civil liberties, environmental concerns, privacy, and security. Regulators globally are also taking steps to address AI risks, with the European Union finalizing laws to regulate high-risk AI technologies and enforce transparency in data usage for AI training.
It is crucial for the United States to establish a regulatory framework that scrutinizes the widespread deployment of AI systems in various domains like transportation, education, and workplaces. AI companies that flout regulations should face consequences. The concentration of AI development within a few tech giants raises concerns that these firms may monopolize expertise in AI, influencing regulatory decisions to their advantage.
Citizens and policymakers must actively engage in shaping the discourse around AI usage, not solely focusing on how or when AI is deployed but also questioning whether certain AI applications should exist at all. Red lines should be drawn to prohibit harmful practices like using AI for predictive policing or basing employment decisions on unreliable emotion-recognition systems.
Empowering the public to demand transparency, independent evaluation of new technologies, and defining limits on AI development is essential. This shift in the AI policy agenda from futuristic threats to immediate concerns is necessary to hold companies accountable and safeguard public interests. The dialogue on AI regulation should be inclusive, drawing insights from diverse communities and labor movements to ensure a balanced and ethical approach to AI governance.
