2.3 Critical Issues in GenAI
While the generative pretrained transformer (GPT) upon which the ChatGPT model is based is not new, OpenAI’s launch of ChatGPT in November 2022 marked the fastest recorded adoption of a technology tool to date. In October 2023, the site had 1.7 billion visits in one month, marking the highest level of usage of any application (Carr, 2023). The rapid proliferation of tools and advancements in the technology saw over 100 leaders in AI technology write an open letter urging a collective pause on AI developments more powerful than GPT 4 to give time for security and safety features to develop and for the creation of regulations and governance structures. The need for such regulation or governance extends to full nations, organizations as well as science and education.
The innovation and creativity in the area of genAI are exciting. These applications can assist with ideation and writing, video and presentation production, research on small and large scales, data analysis and interpretation, image generation, music production and more. ChatGPT can also interact with humans verbally and via camera in such as way as to mimic natural human interactions. However, these systems do not come without limitations or ethical challenges. Some of these challenges speak to the specifics of academic contexts – like academic integrity – while others intersect with communities, organizations, governments, the environment, and humanity as a whole. Broader issues related to genAI include privacy of personal data, risks of misinformation, existential risks, concerns about job dislocation or loss, environmental costs, labour exploitation, and copyright. And specific to the technology, many AI experts have documented alarming concerns relating to the size and scale of large language models, misinformation, AI misalignment, and existential risks to humanity.
Whereas we may, as typical users, have little if any control over how the models are trained, what data is used for training, and how the algorithms process the data, we nonetheless have some control over how we use the models, their applications, and their output. As the humans in the process, we have the responsibility to maintain agency in the human-AI interaction, and we have the obligation to engage in the process by applying due diligence, ethical standards, and a critical perspective. And the way to embrace this human agency is to develop literacy about genAI and apply sound principles for use.
The ethical issues surrounding the development, operation, use, and maintenance of genAI models and applications are numerous and complex. This chapter offers an overview of key areas of concern; for more detailed discussions of the topics, please refer to the References, which provides additional resources for consideration.
Bias
GenAI tools are trained on a range of data, and some frontier models like GPT and other large language models (LLMs) were trained on a wide range of sources includin