A recent analysis highlights constitutional democracy as an essential, yet often missing, element in discussions surrounding AI safety. While the potential risks of artificial intelligence, such as cyberattacks and hyper-persuasion, are now widely debated, the role of political freedom has not been a central focus.
The argument posits that regardless of how technically 'safe' an AI system is designed to be, its safety for citizens is compromised under authoritarian rule. In such contexts, AI could be used for government surveillance, censorship, propaganda, and political suppression.
Conversely, within a healthy democracy, AI could offer significant benefits, including advancements in science and medicine, expanded education, and bolstered economic prosperity. It could also potentially strengthen democracy by helping citizens engage and improving government accountability.
The analysis also suggests that AI itself could challenge democratic institutions. Potential economic insecurity resulting from AI's impact on jobs might create opportunities for demagogues. Furthermore, AI could increase instances where governments invoke emergency powers, and its integration into executive functions might disturb traditional checks and balances, raising the risk of political surveillance.
To counter these risks, the analysis proposes updating legal codes to limit executive power, reforming emergency and surveillance laws, and ensuring due process. It also suggests equipping legislative and judicial bodies with AI tools to maintain oversight as executive agencies leverage AI.