This is a scripted chat. The conversation, the car, the insurer (Hearth) and the prices are invented, and there is no real pricing or payment.
Decision Coverage is a speculative concept for insurance products that cover deviating from AI system recommendations. It is based on the scenario "The Burden of Deviation". Mozilla is not likely to offer insurance. The purpose of the concept is to question the implications of AI decision-making. It accepts the shift in who has to justify a choice and puts a price on it, and it takes no side on whether that is a good thing. The prototype is a short scripted chat on a phone-sized screen. After someone chooses against the assistant's recommendation of a used car, an in-chat sponsored card offers a plan built from the conversation.
As AI systems become more reliable, following their recommendations becomes the default, so choosing against them starts to need a justification that following them does not.
Anyone who relies on AI systems for decisions has this problem, from everyday conveniences such as autocomplete, navigation and shopping to more consequential choices about careers, relationships, family, health and care. These systems hold extraordinary amounts of information about a person, and they may regularly produce better outcomes than the person would alone. Over time, the social expectation changes.
It matters because the person who chooses against the system may be the one who has to justify the choice and bear the cost if it goes wrong.
Instead of arguing about whether people should follow AI recommendations, the prototype accepts that the burden of justifying a choice has shifted and puts a price on it. The idea is shown as an ad inside the chat, the place where the recommendation is made.
After the person chooses the Civic over the assistant's Corolla, a sponsored card appears: "Made up your own mind? Keep yourself safe." It leads to a plan built from the conversation, such as a rental car, a longer return window, a hold on the Corolla and an optional repair add-on. The page says the price is set against the assistant's own forecast. It is a speculative scenario, and the prototype takes no side.
The core hypothesis underlying Decision Coverage is that AI-supported decision-making and recommendations become a default as people become increasingly familiar with these systems and have experiential evidence of their proficiency. We see this in programming and in emerging studies of acceptance rates of AI-suggested work among experts.
If that holds, the assistant's recommendation becomes the baseline that cost and blame are measured against. Choosing differently is still allowed, but it comes with a priced policy, and the price is set by how far you stray from what the assistant predicted.