Shaping the Future of Generative AI for Black Communities: A Frame Analysis of Public Discourse and Empirical Scholarly Research

Imagine a world where computers can write stories, generate images, and answer questions with the ease of a human conversation. This technology, known as generative artificial intelligence, is rapidly moving from science fiction into schools, hospitals, and workplaces. But as these tools become part of daily life, a critical question arises: who do they serve, and who do they leave behind? For decades, researchers have known that computer systems can learn the prejudices of their creators, often treating people of color unfairly. This new study asks a deeper question about how we talk about these problems. It looks at two different groups of people who are trying to fix these issues: the journalists and community leaders writing in the news, and the scientists publishing papers in academic journals. The researchers wanted to see if these two groups were looking at the same problem in the same way, or if they were speaking different languages entirely.

The study, conducted by a team of scholars and industry researchers, set out to compare how the public understands the dangers of artificial intelligence for Black communities versus how scientists study those same dangers. They gathered two large collections of text to analyze. The first collection consisted of 28 public resources, such as news articles, opinion pieces, and reports from think tanks, which represent how the general public and media discuss these issues. The second collection was a systematic review of 91 scientific papers published in major computer science conferences between 2010 and 2025. The team used a method called frame analysis to read these texts. This approach does not just count words; it looks at how a story is constructed. It asks four specific questions: How is the problem defined? What is identified as the cause? What is the moral judgment of the situation? And what solution is proposed? By applying this lens to both the news and the science, the researchers could map out exactly where the two worlds align and where they drift apart.

The findings reveal a significant disconnect between how the public sees the issue and how researchers are trying to solve it. In the public discourse, the story is clear and structural. When journalists and community leaders discuss harm caused by artificial intelligence, they almost always point to deep, historical roots. They describe the problem as a result of centuries of systemic racism and unfair policies that have shaped society. Their proposed solutions are equally broad, calling for government regulations, changes in hiring practices, and new laws to protect communities. They see the technology as a mirror reflecting a broken society, and they argue that fixing the mirror requires fixing the society itself.

In contrast, the scientific literature tells a different, much narrower story. The researchers found that the vast majority of the 91 academic papers focused almost exclusively on technical fixes. When scientists identified a problem, they rarely looked at history or society as the cause. Instead, they pointed to the data used to train the computer models. They argued that the bias existed because the training data was flawed. Consequently, their solutions were almost entirely technical: they proposed creating better datasets, writing new code to detect errors, or adjusting the algorithms. While the public discourse called for laws and societal change, the scientific papers called for better engineering. The study suggests that this is not an accident or a simple difference of opinion. Rather, it indicates that the scientific community has largely stopped short of addressing the deeper structural conditions that allow bias to exist in the first place.

Perhaps the most striking discovery was what both groups agreed on, and what they both missed. Both the news articles and the scientific papers agreed that the most visible harm was "representational harm." This refers to the way these systems distort or erase Black identity, such as generating images of Black people with lighter skin or using stereotypes in text. Because both groups focused so heavily on this specific type of harm, they both proposed solutions aimed at making the images and words more accurate. However, in doing so, both groups largely ignored a crucial perspective: the idea that Black communities should be the ones leading the design and decision-making of these technologies. The study found that in the news, Black people were often portrayed as victims needing protection, while in the scientific papers, they were treated as subjects to be studied or data points to be measured. Neither the journalists nor the scientists consistently framed Black communities as the experts or the architects who should decide how these powerful tools are built and governed.

The researchers argue that this shared focus on representation, while important, is a trap. By concentrating only on making the technology look fairer or more accurate, both groups are missing the bigger picture. They are trying to fix the output of the machine without questioning the rules that govern the machine's creation. The study suggests that true progress requires more than just better data or new laws; it requires a fundamental shift in who holds the power. It calls for a future where Black communities are not just consulted as a formality, but are recognized as the primary authorities on what these technologies should do and how they should be used. Until the scientific community moves beyond technical tweaks to engage with the structural realities of society, and until the public discourse moves beyond seeing Black people as passive recipients of technology, the gap between the promise of artificial intelligence and the reality of its impact will remain wide. The study concludes that bridging this gap requires a new way of thinking, one that treats Black knowledge and experience not as a variable to be corrected, but as the foundation for building a better future.