This study looks at how higher education faculty are contributing their content and services to AI applications, with separate data sets for contributions to commercial AI companies, university applications, trade and professional association efforts, and more information efforts among academic peers. The study also examines how faculty view contributing to AI applications, and whether they believe that their content may have already been used without their consent. The study measures paid sales or licenses to AI applications, as well as gratis contributions.
The study also looks at the issue of the development of libraries or archives of prompts for ChatGPT, Bard and other applications, and defines the percentage of faculty participating in such activities. AI applications perform only as well as their end users enable them to, and a new emerging form of intellectual property – AI application prompts – are a fertile area for department chair, research offices, teaching and learning centers, institutional digital repositories and academic libraries.
Just a few of this 129-page report’s many findings are that:
-
3.6% of faculty were sure that their content was being used by AI applications without their consent and another 38.61% were unsure.
-
Male faculty were far more likely than female faculty to have taken part in efforts to develop AI applications.
-
Efforts to develop archives or libraries of prompts for ChatGPT were most common among faculty in medicine, mathematics/statistics/computer science and architecture/fine/visual arts.
Currently, the major AI platforms are engaged in an arms race to build the biggest databases, the ones trained on the most content. However, industry observers often suggest that the AI industry will turn from what one might call quantity to quality. Applications will be focused on particular applications, and appeal to a narrower, but paying audience. The scientific databases of the future will not only be places to find information, but to manipulate information, develop new information based on old information, and even conduct science. This potential may be a game changer for those entrusted with archiving or safeguarding university assets in science and technology.
Data in the report is based on a representative survey of 777 higher education faculty; the data is presented in the aggregate and also broken out by a broad range of institutional and personal characteristics including age, gender, race/ethnicity, income level, work title and academic field, as well as institution size, type and public/private status, among other variables.