We're building a dystopia just to make people click on ads

In the final weeks before a national election, a voter opens her phone to find a stream of political videos and ads that perfectly match her past searches and likes—some warning of hidden threats from one candidate, others celebrating another. These messages arrive not from official campaigns but through the invisible sorting of platforms that prioritize engagement above all.

Tufekci's Central Claim

Zeynep Tufekci's talk centers on the warning that we are building an artificial intelligence-powered dystopia one click at a time. The same algorithms companies like Facebook, Google, and Amazon deploy to maximize ad clicks also organize users' access to political and social information. This shared infrastructure manipulates behavior at scale, fostering misinformation and polarization as side effects of commercial optimization.

Tufekci grounds her argument in the observation that the technical systems built for advertising revenue have been repurposed without public oversight. Because these tools already excel at predicting and shaping what captures attention, they readily steer political content toward the most reactive audiences, amplifying divisions rather than informing citizens.

Applying the Argument to Election Integrity

The opening scenario illustrates Tufekci's point precisely. When platforms treat political messages like ads—optimizing for clicks rather than accuracy—voters encounter curated realities that deepen distrust. Targeted delivery and the rapid spread of misleading material erode the shared factual ground elections require, turning the infrastructure of commerce into a vector for democratic erosion.

Toward Policy Responses

Tufekci's analysis underscores that any safeguard for elections must address the underlying ad-driven architecture rather than merely patching individual harms. The central question that follows is how policymakers can regulate these AI systems to protect election integrity without eliminating the innovation that drives broader technological progress. Her framing makes clear that the problem is structural, not incidental, and therefore demands structural attention.

What design choices will we accept when the same code that sells shoes also decides what citizens know about their leaders?