A failure of the AI consciousness discourse is that people are talking past each other. Opponents of taking AI consciousness seriously focus on arguing against computational functionalism and related theories. But this is not a response to the proponents. The case in favor of taking AI consciousness seriously is not that computational functionalism or related theories are correct, or that AI systems on current architectures are or will be conscious. Most people who work on the topic are uncertain about these claims. The case is instead that, first, whether AI systems are conscious is an important issue. If we get it wrong in either direction, either by treating conscious systems as objects or by treating nonconscious systems as subjects, we could make serious mistakes. Second, this is also a difficult issue. It touches some of the hardest problems in science and philosophy, and humans have been badly mistaken about nonhuman consciousness in the past, as our shameful history of underestimating animal minds illustrates. Together, these points mean that this issue calls for caution and humility. Rather than confidently affirming or denying AI consciousness, we should invest in evidence-based research and policy that can help us avoid both under- and over-attribution. Moreover, we should develop this work with not only current but also future systems in mind. AI development is moving fast, and future systems may have greater capabilities, greater autonomy, and more biology-inspired architectures than current ones. We should prepare for the possible emergence of such systems now, rather than be caught flat-footed if and when they arrive. Rejecting this case thus requires more than showing that consciousness on current architectures is unlikely. It requires showing that AI consciousness on current *and* future architectures is *so* unlikely that we should neglect the topic entirely, notwithstanding its importance and difficulty. This level of confidence is simply not warranted in either direction at this early stage. To be clear, this is not a response to people who are merely skeptical that AI systems on current architectures are or will be conscious. There are good arguments for and against a range of theories, and I share this skepticism myself. It is instead a response to people who dismiss the topic altogether, and who disparage those who take it seriously as naive or as zealots. Here are a few resources that discuss these issues at a high level: aeon.co/essays/an-ant-is-dro… ai-frontiers.org/articles/a-… youtube.com/watch?v=yEfvhjuj… And here are a few resources that discuss these issues in more detail: link.springer.com/article/10… arxiv.org/abs/2411.00986 nonhumanminds.org/studying-a…