Imagine the quiet, internal work of dreaming up a future for yourself. It is not just a daydream; it is the mental process of gathering the tools, stories, and images we need to picture who we might become. We do not invent these possibilities in a vacuum. Instead, we borrow them from the world around us: from the books we read, the movies we watch, the advertisements we see, and the conversations we have. These shared cultural materials act as a kind of mental scaffolding, helping us construct ideas about what is possible, what is desirable, and what is even worth striving for. When these materials are rich and varied, our imagination has room to grow. When they are narrow or distorted, our vision of the future shrinks. This is the realm of what researchers call "aspirational affordances"—a term that simply describes how the resources available to us shape our ability to reach for new goals.

Now, consider how artificial intelligence is rapidly changing the way these resources are made and shared. AI systems are no longer just tools for sorting data; they are becoming the authors of our stories, the curators of our images, and the guides for our personal growth. They write career advice, generate images of the future, and offer companionship. As these systems permeate our lives, a critical question arises: what happens to our ability to imagine a different future when the very tools we use to dream are controlled by a few powerful algorithms? This is the central concern of a new paper by Sina Fazelpour and Meica Magnani from Northeastern University. They argue that we need to look beyond the usual worries about AI, such as whether it treats people fairly or distributes resources correctly, and instead examine how it might quietly limit what we believe is possible for ourselves and our societies.

The authors introduce a specific concept to explain this phenomenon: aspirational affordances. Think of these as the mental "handles" or "grip points" that culture provides, allowing us to grab onto a new idea of who we could be. Just as a physical staircase affords climbing to someone with the ability to climb, a cultural story or image affords a certain kind of aspiration to someone who can understand it. These affordances are not deterministic; they do not force us to act in a specific way. However, they do make some paths feel obvious and attainable while making others feel invisible or impossible. The researchers suggest that when AI systems generate these stories and images, they are not just adding more content to the mix; they are fundamentally altering the landscape of what we can imagine.

The paper outlines three distinct reasons why the influence of AI on our aspirations is different from, and more dangerous than, traditional media. First, AI interactions are often private, personalized, and conversational. Unlike a movie or a book that is shared publicly, an AI chatbot might act as a life coach or a therapist, offering tailored advice in a one-on-one setting. This intimacy can make the suggestions feel more objective and harder to question. Second, the influence is ecological, meaning it changes the entire environment in which ideas are created. AI is now used to write scripts, design advertisements, and plan cities. This means the very pool of ideas that humans draw from is being filtered and reshaped by algorithms before it ever reaches us. Third, this power is highly concentrated. A small number of corporations control the major AI systems that generate these cultural resources. When only a few voices dictate the stories of the future, the range of possibilities narrows, and alternative visions risk being erased.

To make sense of these risks, the authors propose a new category of harm called "aspirational harm." This is distinct from the more familiar harms of AI, such as representational harm (where a group is portrayed in a stereotypical or insulting way) or allocative harm (where resources are distributed unfairly). Aspirational harm occurs when AI systems distort or diminish the available interpretive resources in a way that undermines a person's ability to imagine practical possibilities for their life. It is not necessarily about being insulted or denied a job; it is about being unable to see a path forward. Even if an AI system portrays a group with respect, it can still cause aspirational harm if it only shows them in a narrow set of roles, effectively closing the door on other ways of being.

The researchers illustrate this with three concrete examples. In the first, they look at career guidance. An AI system might try to help women enter science and technology fields by showing them images of successful female scientists. However, if the system only presents a single, rigid archetype of success—perhaps one that requires extreme resilience in hostile environments—it might inadvertently erase other valid ways of being a scientist. It replaces a diverse range of possibilities with a single, narrow script, limiting the imagination of young women who might have wanted to pursue a different kind of career path.

In the second example, the authors examine how AI visualizes the future. They tested this by asking an image-generating AI to create a picture of Iranian schoolgirls in the year 2040. The real-world context involved a movement for political freedom and the removal of mandatory dress codes. The AI, however, generated images that focused entirely on technological advancements, such as holographic tablets and eco-friendly architecture, while keeping the social and political status quo completely intact. The girls were still wearing hijabs, and the political restrictions remained. The AI had the power to imagine a future, but it defaulted to a version where technology changed but human rights did not. This effectively erased the aspirations of the protesters, making their vision of a free future invisible and, by extension, less plausible to those who see the images.

The third example concerns how AI shapes our understanding of what is socially possible. When AI systems are used to summarize news or answer questions, they can subtly steer users toward dominant viewpoints while suppressing alternative narratives. If an AI consistently presents the current social order as the only logical or inevitable outcome, it can discourage people from imagining that things could be different. This is not just a matter of bias; it is a matter of shrinking the mental space available for social change. By making the status quo feel like the only reality, these systems can erode the hope necessary for transformation.

The paper concludes that these harms are real and require urgent attention. The authors are not suggesting that AI is a monster that will steal our dreams, nor are they claiming that the damage is inevitable. Instead, they argue that we must recognize the unique power of these systems to shape the very resources we use to think about our futures. The danger lies in the quiet, cumulative effect of millions of personalized interactions and algorithmic choices that slowly narrow the horizon of what we believe is possible. By identifying "aspirational harm" as a distinct category, the researchers hope to open a new conversation about how to design AI systems that expand our imagination rather than confine it. The goal is to ensure that as these technologies become more integrated into our lives, they continue to serve as tools for resistance and transformation, helping us to envision and build a future that is truly open to all.