The US government's reliance on artificial intelligence has reached unprecedented levels, with new figures revealing a 70% surge in active or planned AI use cases across federal departments. The total now stands at an astonishing 3,611 instances, sparking fears about the implications for civil liberties and accountability.
This rapid expansion of AI adoption has raised concerns among experts about the transfer of critical decision-making processes from human oversight to automated systems. Highly sensitive functions such as individual freedoms, public health and wellbeing, and nuclear reactor safety are all set to be integrated with AI. Critics argue that the absence of robust public scrutiny and transparency is a major red flag.
Take, for example, Palantir's provision of AI software to scrutinise grant applications for ideological alignment at the Health and Human Services (HHS) office. The Federal Bureau of Prisons is also developing an AI system to predict potential misconduct in newly admitted inmates – which could influence security classifications before any wrongdoing occurs.
Even more worrying are the Department of Veterans Affairs' plans to deploy AI to monitor calls to its crisis line, gathering external data on mental state and suicide risk. Meanwhile, the Department of Energy is exploring AI for controlling nuclear reactors and autonomously responding to safety incidents – raising serious questions about accountability and the potential risks.
While some applications may have beneficial outcomes, such as using AI to assist vulnerable veterans, the lack of transparency surrounding these deployments is a major concern. The Office of Management and Budget's (OMB) disclosures provide woefully inadequate information, often just a single sentence, offering little context about the systems' purpose, methodology, or safeguards.
Public consultation on these AI programmes appears to be largely non-existent. Despite policies that theoretically allow for public involvement, engagement is rare – making it difficult for citizens to scrutinise or provide input on systems that could profoundly affect their lives. This lack of transparency and public engagement raises significant questions about accountability and the ethical governance of AI within government operations.