The Present and Future of Online Discourse

The talk opens with the speaker's account of building consumer and political technology in the early 2010s, including crowdsourcing and campaign tools. Each project had safety reviews and ethics discussions, yet the same mechanisms were later used for disinformation in ways the team had not anticipated. The speaker presents this as a recurring failure: builders focus on one side of a powerful tool and rarely work through how it could be misused.

The middle of the talk turns to AI and the labor market. The speaker describes tech companies cutting mid-level engineers while keeping senior coaches and low-cost junior staff, and suggests that revenue-per-employee figures are an early signal of where cuts will land. To counter these risks he recommends diverse teams, tabletop exercises that bring in people from outside engineering, failure-mode thinking borrowed from aerospace, and asking whether a system should be built, not only whether it can be. He also argues that verifying humanity may matter more than detecting machine-generated text, and that countering disinformation is hard once facts become political.

In the question period the talk moves to hiring, apprenticeships, teaching students about ethics, and whether online communities can recover from their current state. The speaker's takeaway is that technologists should shape the future they want rather than leave it to others, and that smaller trusted spaces may help people step back from constant online performance.