Why Equity Must Be a Core Part of the Conversation About AI

• 08/06/24

There is also a broader digital literacy divide to consider. Users' experiences with AI will differ based on their familiarity and comfort with technology as a whole. A significant portion of the population, particularly those in jobs that do not require consistent computer or internet access, may not be accustomed to using these technologies. Transitioning from the infrequent use of search tools like Google and other online resources to effectively using AI can be daunting, requiring a longer learning curve. This disparity underscores the need to acknowledge that not everyone is starting from the same level of understanding and technological proficiency.

Importance of Designing AI with Equity from the Start

It is ideal to design AI products with equity from the start. Retrofitting equity considerations into an already developed product is problematic, as it often fails to address foundational biases and systemic inequities. Equitable design ensures that diverse needs and potential disparities are integral to the development process, leading to more inclusive and effective solutions. Without this proactive approach, AI systems are likely to perpetuate or even exacerbate existing inequities.

In the absence of tools built with equity from the onset, the best alternative approach is to do our due diligence in finding the best available technology. This includes thoroughly testing existing software for biases, involving a diverse group of individuals to provide feedback, and evaluating the software's performance in different contexts. Once we identify suitable tools, being transparent about how the AI is used is essential. Transparency in AI usage helps build trust and allows users to understand the potential biases and limitations of the technology.

As we discover more about the technology, we must regularly update the algorithm to incorporate new information, correct biases, and improve accuracy. By continuously refining the AI with diverse and inclusive data, conducting ongoing testing, and integrating feedback from a wide range of users, we can help the algorithm evolve to better serve all users equitably.

Additional Considerations and Challenges

It's important to note that AI is not a panacea. While it holds great potential to "equitize" education, its use can also pose risks if equity is not a central consideration in its creation and implementation. Challenges such as bias in AI algorithms, data privacy concerns, and the digital divide must be addressed to ensure AI benefits all students. Additionally, there is the potential for over-reliance on AI, which could hinder students' growth in certain areas if they become too dependent on these tools. Moreover, AI-generated outputs can sometimes fail to reflect the user's unique voice, potentially leading to a lack of authenticity in students' work.

Moving Forward

Given these challenges, education industry leaders and ed tech providers must actively ensure that AI can be accessed and implemented equitably. One area of progress in recent years has been an uptick in state and federal projects that aim to increase broadband coverage for schools and rural residents. Lawmakers from Kentucky to Alaska have set aside billions of dollars in funding to improve high-speed internet access for schools and communities. Education interest groups, activists, and higher ed institutions have heavily advocated for these bills, and their success demonstrates the power and necessity of industry-wide collaboration. Key stakeholders must continue pushing for public policies that address the digital divide.

AI equity is top-of-mind for my work at Lumen Learning, where we are exploring methods to address AI disparities from our unique position as courseware developers. Though we are still in the early stages, my colleagues and I have prioritized a thoughtful and intentional approach as we have begun to grapple with the technology and its implications for our work. Several key considerations drive this strategy.

Rather than implementing AI for the sake of adoption, ed tech providers should focus on two things: 1) thoughtful integration over hasty deployment that could potentially lead to issues related to bias, accessibility, and overall effectiveness, and 2) how AI can be strategically leveraged to improve educational outcomes and empower both students and faculty members. While we are dedicated to fostering equity and inclusion within our educational products, this commitment must include ensuring equitable access to AI resources within our courseware.

By focusing on these areas, we can create an educational environment where all students, regardless of their background or prior exposure to technology, can benefit from AI advancements.