When a disaster is averted, the public often assumes the threat was never real. This is the Preparedness Paradox, and it has done immense damage to institutional trust. The Y2K bug is the classic example. Programmers worked for years to prevent a global systems collapse. When nothing happened, millions concluded the whole thing was a hoax. The invisible effort became, in the public eye, evidence of a lie.
This pattern repeats across crises. During the early days of COVID-19, conflicting guidelines eroded public faith in health authorities. The disconnect stemmed from a failure to convey uncertainty. Experts knew little at the start, but the public, viewing events with hindsight, believed they knew everything all along. The result is a toxic cycle of blame and distrust.
Why Institutions Cannot Fix This Alone
The world, particularly the United States, is now grappling with unprecedentedly low levels of trust in authority and institutions. This historical baggage is a critical problem for AI safety. AI is a field dominated by organizations like Anthropic and OpenAI. While these labs do discuss existential risks, their advocacy is often perceived by the public as marketing designed to attract investors or regulatory capture, rather than genuine concern. When the only voices warning of extinction-level risk are the companies building the technology, skepticism is a rational response.
This is an inherited crisis of credibility. Trust is typically rebuilt through many small disaster-response cycles. An institution handles a crisis well, proves its competence, and slowly regains faith. AI safety may not have the time for that. The timeline from today to the creation of AGI is unknown, but it is short enough that we cannot wait for trust to be rebuilt through traditional means. The public is whiplashed after years of ineffective communication across all sectors. If AI is potentially going to result in a global catastrophe, we need messengers the public will actually hear.
The Power of the Individual
The solution to this institutional trust deficit is decentralization. It will likely be up to individuals rather than institutions to persuade the public about AI risk. Individuals are imperfect messengers, certainly. They are prone to ad hominem attacks and unstable. However, they are also relatable. A person on a podcast or a blog speaking with conviction about the risks of AI carries a weight that a corporate press release never can.
There is currently a large portion of Americans who are against AI altogether, but there isn't much bottom-up support for AI safety specifically. The movement needs messengers who can bridge the gap between "AI is cool" and "AI will kill us." We need individuals who can translate abstract technical concepts into concrete, understandable language for a general population. The goal isn't to sound the alarm bells in a way that makes people ignore them, but to communicate the risks and the importance of this issue effectively.
The Path Forward
We are inheriting decades of eroded trust. Relying solely on tech executives to fix this is a losing strategy. The public is searching for authentic voices. The AI safety movement needs to invest in supporting communicators who are not beholden to the balance sheets of the labs they are critiquing. This is a worthwhile lever to pursue. The future of the world may depend on whether we can successfully communicate what is coming, before it is too late.