AI ethics: Protecting privacy

- It's been a whirlwind morning for Aaliyah, HR manager at Every Company. She's been in a meeting with a new vendor to help fast track the development of an in-house chat bot, which will use generative AI to support the HR department. This in-house generative AI solution is a high priority because Aaliyah has learned from a trusted source that some members of the HR team are using ChatGPT to help with their work, including writing offer letters to candidates. This source claims that the data shared with ChatGPT includes transcripts from the interview, the candidate's resume, and sensitive details about the offer, such as starting salary. Aaliyah is alarmed by these claims and what this means for data privacy. She also wonders how else ChatGPT is being used and what other sensitive employee data is being shared. Since every company does not have a formal, corporate-wide policy about the use of AI tools, it makes the job of managers like Aaliyah more challenging because every department is left to try to enable their own standards, and it's been confusing for employees. For example, the marketing department is actively encouraging generative AI use, and they are fine with their team using free tools like ChatGPT. Aaliyah is right to be concerned about the rumors of HR staff using ChatGPT for personalized correspondence. Recall our discussion about privacy and the legal and ethical implications of sharing personal or sensitive data with a generative AI tool. Imagine the personal details that would need to go into a job offer letter, everything from a person's name and address to the salary and job start date. To make the letter truly personalized, it seems like interview and resume details are being used as well without the candidate's consent. There could be ways to mitigate some of the risks. For example, having ChatGPT just provide an outline of generic information for the job title and then update the letter with all the sensitive or personal details. However, at that point, it might make more sense to just use a template rather than generative AI. Aaliyah calls an HR department meeting. She acknowledges that generative AI can be a powerful tool, but she also points out the risks and asks her team to not use free tools like ChatGPT and instead to wait for the in-house solution that will offer greater risk mitigation for every company to use generative AI in HR. Remember the in-house generative AI solution for the HR team will have more guardrails in place. It will be implemented behind the corporate firewall for greater cybersecurity, mapped to corporate data sources to provide greater accuracy, and the legal terms will include clarity around data privacy matters. In the meantime, she signs up the team for some introductory training on responsible AI use so that everyone in HR understands the limitations of AI tools and will be better equipped to use generative AI once the official solution is rolled out. Aaliyah also plans to speak with her VP to advocate for a company-wide policy on AI governance. This will help provide a clear and consistent set of guidelines for everyone, especially when you're working with sensitive information.