OpenAI's ChatGPT app is displayed on an iPhone in New York, May 18, 2023. With companies deploying artificial intelligence to every corner of society, state lawmakers are playing catch-up with the first major proposals to reign in AI's penchant for discrimination — but those bills face blistering headwinds from every direction.
Artificial intelligence (AI) plays a significant role in decision-making, affecting job interviews, housing, and healthcare. However, efforts to address bias in AI decision-making face challenges. Lawmakers from states like Colorado, Connecticut, and Texas are proposing bills to address this issue, but they encounter resistance from various stakeholders.
Civil rights groups and labor unions advocate for more transparency and legal recourse against AI discrimination. However, the industry is cautious about accountability measures. Bipartisan lawmakers from states including Alaska, Georgia, and Virginia stress the importance of collaboration and compromise in crafting AI legislation to avoid regulatory inconsistencies.
While over 400 AI-related bills are under debate nationwide, most target specific industries or technologies. The proposed bills aim to oversee AI discrimination, a prevalent issue highlighted by cases like inaccurate assessments of Black medical patients and gender bias in job applications.
Despite the widespread use of AI in hiring (up to 83% of employers utilize algorithms), bias remains a concern. Proposals in Colorado and Connecticut require companies to conduct "impact assessments" for AI systems involved in decision-making. These assessments would include descriptions of AI's role, data collection, discrimination risks analysis, and safeguards.
While these measures enhance accountability and public safety, companies fear increased legal risks and disclosure of trade secrets. David Edmonson from TechNet emphasizes collaboration to address AI risks while fostering innovation.
Under the proposed bills, companies aren't mandated to routinely submit impact assessments to the government but must disclose discrimination findings to the attorney general. However, labor unions and academics worry that relying on self-reporting may undermine public and government oversight.
The limitation of lawsuit filing to state attorneys general, excluding citizens, is another contentious issue. Workday endorses proposals without citizen lawsuit provisions, arguing that judicial decisions may lack technical expertise, leading to inconsistent regulation.
Sorelle Friedler from Haverford College underscores the importance of citizen lawsuits as a means to assert rights in American society. Despite industry opposition, lawmakers like Connecticut's Sen. James Maroney are determined to create regulations ensuring safe and trustworthy AI.
Maroney refutes claims of industry influence, emphasizing collaboration with diverse stakeholders. Despite challenges, efforts to regulate AI aim to address bias and ensure fairness in decision-making, reflecting ongoing debates about the intersection of technology and society.
