With great power: AI innovation, DEI, and responsibility 

Learnosity EVP John Kleeman, on the importance of responsibility in a fast-moving AI world, why DEI is a double-edged sword, and the transformative potential of AI-powered learning.

With AI quickly jumping from ‘new’ to ‘widely used’ across all industries, the question of responsibility is becoming more important. From a practical standpoint, what do you think it means to apply AI responsibly?

Responsible and ethical AI is a multifaceted challenge, and attempting to use AI in assessment without human supervision is unwise. Often, the essential element is to have a human in the loop at multiple stages of assessment development to identify mistakes and mitigate risk. Doing so complements AI limitations in judgment and discretion and ensures broader social and moral concerns are considered. 

A good resource for assessment organizations thinking about responsible AI is the recent ATP publication: ‘Creating Responsible and Ethical AI Policies for Assessment Organizations’ which sets out 28 topics that an assessment organization should consider when defining their own policies around the responsible use of AI. These topics include governance or internal documentation of AI and the more general issues of cybersecurity and ensuring AI is human-driven. Not every organization will have to consider all 28 topics, of course, but using AI well typically reaches into most aspects of an organization’s structure.

The road to responsible AI should be seen as a journey, not a destination…It’s only by having that hands-on experience with AI tools that you truly gain a good understanding of the possibilities and limitations. Share on X

In short, the road to responsible AI should be seen as a journey, not a destination. To that end, the most important thing an organization can do is encourage its personnel to experiment with AI and learn as much about it as possible. It’s only by having that hands-on experience with AI tools that you truly gain a good understanding of the possibilities and limitations.

Some aspects of the responsibility conversation center on DEI and how AI can potentially amplify bias. As an advocate for baked-in DEI, what’s your take on this?

Many people are reluctant to use AI in assessment due to the potential for AI-generated material to carry bias. That said, the alternative of only using human-created content isn’t one without consequences. Despite ongoing efforts to make human-created assessments fair and inclusive, concerns about their fairness persist, with scenarios such as admissions tests worldwide accused of favoring certain demographics and those with privilege over other candidates. In these situations, where humans can be imperfect creators, AI can be a powerful tool for addressing DEI challenges by helping identify bias in human-created content.

So while both humans and AI can be guilty of bias, I believe that, together, it will be possible to build better assessments than ever before.

Can you tell us more about how the upcoming Item Bank Health Check tool from Learnosity might help support authors with tackling DEI challenges?

Our core aim at Learnosity is to advance education and learning worldwide, which is something we believe AI can be a huge driver of and very much reflects the development of tech like the Item Bank Health Check tool.

With that in mind, one core feature we’re developing as part of this tool is the Bias Analysis feature. In simple terms, this feature reviews a question from a DEI perspective and flags potential concerns. For example, gender-biased language or the use of insensitive language. The big benefit here is not just in the AI’s ability to help authors safeguard DEI, of course, but also that the AI can take the pain out of needing to reframe or recreate the content too.  

…as with all Learnosity tech, we put reliability and responsibility front and center during development to ensure that the end result has a real-world application. Share on X

This is currently being developed by our AI Labs team, based on prototyping and research by myself, our CEO Gavin Cooney, and our CPO Neil McGough. We’re truly excited to see the value this latest innovation can bring to the table and, as with all Learnosity tech, we put reliability and responsibility front and center during development to ensure that the end result has a real-world application.

As you look to the future, what excites you about AI and its potential to transform learning?

The great promise of technology in this industry is that everyone can gain access to the best learning, not just the privileged few. A key reason I came into the field of computerized assessment in the 1980s was to help democratize technology at a time when personal computers were new and not always intuitive to use. I wanted to bring the benefits of easy-to-use PC software to teachers and instructors around the world.

My belief is that with more AI-powered technology at our fingertips, responsibly applied and innovatively used, the learning opportunities we all deserve might finally become accessible to everyone. Share on X

We’ve made huge progress over the years, and there has been a massive improvement in educational attainment in most countries over the decades. That said, there is still a huge amount to do before every individual who wants and needs to learn is given the very best possible means of doing so.

Progress is rarely quick or easy, and what we can do with AI today is only a hint of what might be possible later on. My belief is that with more AI-powered technology at our fingertips, responsibly applied and innovatively used, the learning opportunities we all deserve might finally become accessible to everyone.

To learn more about AI’s impact on assessment, download your copy of our AI guidebook. 👇