The Dark Side of AI: A Professor's Cautionary Tale

As a professor, I've come to rely on AI tools like ChatGPT for my work, but a recent experience has left me questioning their reliability and accountability.

TL;DR
• A professor's 2 years of work was lost due to a ChatGPT data deletion glitch.
• The incident highlights the need for accountability and reliability in AI tools for academic work.
• Researchers should consider the potential consequences of relying on AI tools and take steps to ensure their data is safe.

Summary
The increasing reliance on AI tools like ChatGPT in academia has raised concerns about their reliability and accountability. A professor at the University of Cologne, who used ChatGPT daily for tasks such as writing emails, drafting course descriptions, and analyzing student responses, experienced a significant loss of data when they temporarily disabled the 'data consent' option.

This action resulted in the permanent deletion of two years' worth of work, including project folders and conversations, without any warning or recovery option. The professor had saved partial copies of some conversations but lost large parts of their work forever.

Key issues highlighted by this incident include:
• The lack of protective measures, such as warnings about irreversible deletion and recovery options.
• The absence of backups or redundancy, despite the professor being a paying subscriber.
• The company's prioritization of user privacy over data recovery.

OpenAI, the developer of ChatGPT, attributed the deletion to their 'privacy by design' approach, which ensures that user data is deleted without a trace when the data sharing option is deactivated. The company advised users to maintain personal backups for professional work.

This incident underscores the need for AI tools to be developed with academic standards of reliability and accountability in mind, including safeguards to prevent data loss and ensure user protection.

As we continue to integrate AI tools into our academic work, it's essential that we prioritize accountability and reliability to avoid similar incidents in the future.