"The conversation is sort of happening in Silicon Valley around one thing, and a totally different conversation is happening among consumers." That's Campbell Brown, former head of news partnerships at Meta, speaking at StrictlyVC this week. The disconnect she's naming is real, it's widening, and it has practical consequences for anyone who depends on AI tools.
Two Different Arguments
In Silicon Valley, debates about AI content tend to focus on technical problems: detecting misinformation, filtering harmful outputs, keeping models from saying something that generates regulatory heat. The implicit assumption is that these are engineering challenges with engineering solutions.
Consumers are asking something different. Why did the AI say that and not something else? Who decided what it knows? Is this answer shaped by whoever built the product or trained the model? These are questions about trust and power, not model architecture.
Brown is well-positioned to see this divide. At Meta, she managed the gap between a platform that claimed neutrality and users who experienced its content decisions as anything but neutral. AI assistants are the same tension at higher speed. Decisions that once required editorial discussions now happen at inference time - when the AI generates a response - across billions of queries, with no byline attached and no appeals process.
The Accountability Structure Doesn't Exist Yet
When you ask an AI assistant about a medication, a legal question, or a contested political topic, something upstream determined what the model knows, what it will say, and how confidently it says it. That is an editorial decision, whether or not anyone calls it that.
News organizations have editorial standards. Social platforms have content policies, imperfect as they are. AI assistants operate through system prompts (invisible instructions given to the model before you start a conversation) and RLHF - a training method where human raters teach models which responses are preferred. Neither is visible to users, and neither comes with accountability structures that consumers can evaluate or challenge.
The companies building these tools have strong incentives to keep the conversation technical. Technical problems have product solutions. The question Brown is raising - who is accountable for what AI tells you - has no clean product answer.
As AI assistants replace search engines and reference materials for more everyday decisions, that accountability gap becomes more consequential. Brown doesn't have a solution. But framing the right question is still valuable, and the people who use these tools every day deserve to be part of a conversation that's currently happening without them.