A new AI-driven digital learning platform launched by Arizona State University has set off alarm bells among academics for its exploitation of scholarly work. ASU Atomic is a pilot platform that uses generative AI to allow subscribing users to automatically create bespoke micro-courses out of content previously created by staff for other courses and purposes. The tech news outlet 404media has reported that professors at ASU were “disturbed to find their lectures chopped up and turned into AI slop” without their consent or knowledge.

The case of Atomic has attracted a lot of attention among university staff on social media, many of whom have seen as it as exploitative of academic labour without compensation. One ASU staff member interviewed by Inside Higher Ed called it “Frankensteinian.” It appears that the tool scrapes the content produced by its faculty and uploaded to its Canvas learning management system, then stitches it together into new forms according to subscribing users’ instructions.

It’s notable that ASU has positioned itself in recent years as a highly innovative, tech-friendly and pro-industry institution. It co-hosts the annual ASU-GSV edtech investment summit, has become a major provider of online degrees and courses, and is in active collaboration with OpenAI exploring how “the advanced capabilities of ChatGPT Enterprise” can be used in higher education teaching, learning and research.

ASU Atomic appears to be the next step in ASU’s efforts to innovate with tech in the design and provision of education. It combines the subscription model of online courses with the capacities of AI and automation to produce new revenue-generating educational products and consumer-based leraning services.

More significantly, the Atomic platform also illuminates the growing trend of universities – far beyond ASU alone – exploring how to develop novel income-generating opportunities from their digital archives of staff content. Recently, Janja Komljenovic and I produced a report for Educational International on the challenges that digital platforms pose to academics’ intellectual property and autonomy. In the report, we argued that “platformization” of higher education, even before generative AI appeared on the scene, was already complicating questions about the academic ownership and control of content and teaching materials. AI is now set to amplify the exploitation of academics’ work.

Assetization in the academy

When universities contract with digital learning platforms such as a learning management system provider or an online learning platform, we argued in the report, user content and data can be treated as valuable assets for potential income generation by an educational institution. For example, uploaded content can be used to support further product development, which may then be offered to institutions or individuals for a subscription or similar fee.

Once academics upload content – such as lecture notes, handout or video recordings – then it becomes very difficult to remove from a platform. Moreover, institutions can continue using that content even once a member of staff has left the institution, or even died in service. During the pandemic, HE teachers and researchers became increasingly concerned about HE institutions claiming ownership of teaching materials posted on platforms. Many academics and HE institutions faced new challenges regarding their digital rights over recorded lectures and other materials posted on digital infrastructures and platforms as a result of confusing and contested legal copyright arrangements.

What these developments indicate is the increasing presence of an economic logic in higher education that treats educational materials and data as digital assets with potential financial value. In a more recent article where Janja and I worked with Kean Birch and Klaus Beiter, we described academic content assetization as the ways that scholarly materials are increasingly controlled and capitalized by higher education institutions and digital platform providers. The term digital asset works here because it refers to things that are owned and controlled without being sold or exchanged like a commodity. Assets generate income from subscriptions and fees for access, and are underpinned by complex contractual and copyright arrangements that determine ownership and control rights.

Universities were already involved in academic content assetization through platforms such as massive online open courses (MOOCs) and online program managers (OPMs) a decade or more ago. While OPM and MOOC providers typically offer digital infrastructure and services such as marketing and recruitment, universities contribute brands, reputational status, academic content, and issue certificates and awards, with both benefitting from fee-sharing arrangements when students sign up to the courses. Learning management systems like Canvas and Blackboard has amassed academic content and user data too. These often require academic staff to sign agreements transferring their copyright to their employer, or licencing their content for re-use according to the terms and conditions of each platform.

The digitalization and platformization of HE, we argued, enables the massification and acceleration of academic content creation and access as well as new modes of monetization through the construction of new audiences and, thus, new markets. Two important aspects of academic content assetization are the emergence of content-related derivative digital services and the introduction of large language models into academic knowledge practices. AI firms such as OpenAI have partnered with universities for commercial benefit while also promising institutions that they will benefit from being able to offer new AI-powered services.

We noted in the paper that many universities have sought to change their IP and copyright arrangements so that ownership and control rests with the institution rather than the individual, thus allowing the institution to re-use and re-purpose that materials produced by its own staff for purposes of institutional revenue-generation. Conventionally, universities have not claimed the copyright for teaching materials, but this has changed fast since online learning platforms became new sources of institutional income. It is through such copyright arrangements that teaching content can be configured as a value-making asset for university institutions, as well as a potentially profitable source for any platform providers with which they are in partnership.

Such technical and legal arrangements were already in place through online learning MOOC and OPM platforms, but as ASU Atomic indicates, they may now underpin efforts to capitalize on academic content through new AI services. ASU has an intellectual property policy in place asserting that the institution owns the rights to all teaching materials produced by its staff, for example, and this is the direction that other institutions are heading in too. Academic content and work is now being reconfigured as institutional property and as assets from which future value can be generated in the shape of derivative services and products, even without the explicit consent or involvement of those who produced the material in the first place.

Assetization of academic content therefore raises questions around intellectual property rights and rights to claim economic benefits. It also surfaces challenges related to academic freedom and pedagogic autonomy.

AI and academic freedom

The ASU Atomic platform illustrates how generative AI is likely to amplify the challenges and controversies associated with the assetization of academic content. It demonstrates how, once an institution has claimed ownership of its own educators’ materials, that content can be rapidly re-purposed and re-mixed as “personalized” learning materials available to subscribing individual users.  

The way that Atomic allows teaching materials to be re-mixed for income generation purposes is, additionally, in tension with core values of academic freedom, particularly freedom in teaching. Freedom to teach includes:

Freedom to determine what shall be taught (course content); freedom to determine how it shall be taught (pedagogy); freedom to  determine who shall teach (via transparent selection procedures);  freedom to determine whom shall be taught (the right to determine and enforce entry standards); freedom to determine how students’ progress shall be evaluated (assessment methods); freedom to  determine whether students shall progress (via marking criteria and  grade determination).

The erosion of pedagogic autonomy has long been a problem with digital learning management platforms like Canvas, because they impose templates on how courses can be taught and what content can be used during classes. A platform like Atomic clearly amplifies these challenges over pedagogic freedom. It combines materials into new forms over which the original academic has no control. It replaces educator-led pedagogy with automated teaching provision. And it removes the academic from any direct teaching, evaluation, assessment or progression decisions by relocating all those powers to AI.

Subscriptions-based AI services that re-mix ad re-package academic content as a process of assetization and income-generation, then, exacerbate an erosion of academic autonomy and freedom that has been progressing since platforms become a routine presence in higher education more than a decade ago. It is precisely through such platforms as the Canvas learning management system that academic content has been amassed in forms and quantities that may now be exploited by institutions and companies for economic gain. When institutions claim ownership of content produced by their staff, academic themselves lose control of their own work without any compensation for their labour.

These developments change how academic materials are valued. For educators, the materials they produce are valuable as essential aspects of their pedagogy, and as artefacts of their academic freedom to choose how they teach. From the perspective of an institutional platform such as ASU Atomic, academic content is valuable from an economic rather than educational perspective.

Contractual transparency

While ASU Atomic may be generating anxiety in the sector already, it is unlikely to be the only case of institutions re-using academic content as value-creating assets that promise future economic benefits. Although it is not clear which AI model underpins Atomic (though given ASU’s partnership with OpenAI to construct new applications, we can guess), this is clearly the outcome of a contractual relationship enabling ASU to build on top of an existing model. As Janja and I recommended in our report for Education International, contractual processes between universities and edtech vendors should be made much more transparent so that it is apparent to academic workers what rights they retain over their content when universities contract with platform and AI companies.

Contracts with vendors may seem to be boring legal documents, but they constitute important processes where significant aspects of academic freedom and IP are negotiated. Yet academics are often unaware of copyright arrangements in their institutions, or of the contractual agreements being made with platform vendors, and may not be consulted regarding decisions that will ultimately impact their working conditions, rights or academic freedoms. Now that ASU Atomic has indicated how AI will amplify such problems and exploitative behaviours, by enabling universities to capitalize on the content assets produced by academic staff, educator unions and associations should act fast to protect workers’ rights and autonomy.