ChatGPT Usage Policies: Five Key Considerations for Businesses

The advent of ChatGPT has disrupted the business landscape, presenting both opportunities and challenges to organizations. During my LinkedIn training sessions and in conversations with business professionals across industries, the topic of ChatGPT keeps popping up.

 If you haven’t started experimenting with it yet, ChatGPT is an AI-based chatbot that is able to generate remarkably human-like text responses to almost any question. Our purpose here is not to explore the machine learning algorithms, natural language processing, and other technologies that make ChatGPT possible. Suffice it to say that this technology is rapidly transforming how we work, search, create content, and brainstorm ideas, to name just a few areas. 

 Forward-thinking organizations are strategizing  about how to get ahead of the impact of this technology on their businesses as their employees are already using it in many cases. Some companies have blocked access to ChatGPT on corporate systems to reduce the risks of inaccuracies and other errors. However, anyone can still access ChatGPT through mobile devices and personal computers, creating a need for policies on how employees should use this technology.

Businesses can’t be too restrictive with guidelines as there are limits on what’s enforceable, and they don’t want to throttle potential innovation by team members. One point of reference could be how enterprises created policies for the use of social media - another instance of a technology that rapidly achieved mass adoption, forcing organizations to react to what their employees were already doing.

 As you evaluate which organizational guidelines to establish for the use of ChatGPT by employees, consider these five factors:

1.   Inaccuracies

 ChatGPT sometimes returns inaccurate results, and wrong answers are presented as authoritatively as correct answers. It's important to recognize that ChatGPT doesn’t know when it is wrong, and employees need to be trained on how to verify its responses using other sources. Sometimes these “hallucinations” are obvious, but in other cases they may sound perfectly plausible. Critical thinking skills and common sense are just as important as ever in spite of our tendency to take information provided by automated tools at face value.

 2.   Authenticity and transparency

 ChatGPT can be a powerful tool for content creation, customer service, programming, and many other applications, but it's essential to maintain authenticity and transparency. Processes need to be established to clearly indicate to readers, customers, and others when ChatGPT is a major contributor to any interaction. For better or worse, it won’t be long before everyone will assume AI is involved in most communications. 

 ChatGPT is not a substitute for professional advice: For regulated industries such as healthcare, financial services, and law, ChatGPT should not be used as a source of professional advice. Employees need to understand the risks of using information that may be inaccurate or outdated and understand their professional obligations when interacting with clients.

 A downside of ChatGPT is that, in its current version (GPT-4), it doesn’t provide sources for its answers to questions. This lack of transparency needs to be corrected in future versions of this tool. Bing’s version of a similar tool (aka Sydney), which runs on ChatGPT, does include references in its responses, and there is a Chrome extension (WebChatGPT) that provides some references for ChatGPT responses.

 3.   Ethical Considerations

 An ethical issue arises when an individual publishes content generated by ChatGPT without revising it to reflect their own voice and doesn’t mention that ChatGPT was used. This naturally raises the question of where you draw the line between ChatGPT created content and “original content” when deciding whether to acknowledge the tool as a coauthor. Business policies need to be created to help employees understand how to handle these issues so they don’t have to figure this out on their own.

ChatGPT can also pose challenges with unintentional plagiarism and related areas around intellectual property ownership. On the bright side, GPT-4 cites references for the content it generates which is an improvement over GPT-3.5. It will take some time for our legal and regulatory systems to figure out how to formally address these issues.

 4.   Privacy and confidentiality

ChatGPT queries may be about client questions and issues, raising concerns about privacy and confidentiality. It's essential to establish guidelines for employees to follow, such as not entering any confidential or otherwise sensitive information into the platform.

 5.   Training and Supervision

ChatGPT is a powerful tool that requires proper training for employees to use it effectively within a business context. Businesses should provide education about the features and limitations of ChatGPT, and encourage team members to ask for help whenever they need clarification on specific questions. In addition, feedback loops should be established to enable employees to report issues and make suggestions for improvements.  

ChatGPT has introduced a new dimension to the world of human-machine collaboration and many organizations are already establishing guidelines for its use. With proper training and guidelines in place, employees can leverage this technology to enhance their productivity and efficiency while minimizing the risks associated with its use.

 As you can imagine, this article was produced in collaboration with ChatGPT.