Real-World AI Use Cases at TXST

I use ChatGPT for code generation not because I don't know how to code, but because it is much faster. ChatGPT is a fantastic "ghost writer". I can prompt it with a request and if I think its response contains mistakes, I can tell it why and then ask it to give me another response. It will find its own mistake and most of the time the next response is a good answer. If it makes another mistake, I can repeat the process until I get a response that is worth attempting to deploy to my application. I think of my relationship to ChatGPT as a collaboration.

I used ChatGPT to collect information from multiple websites that would have taken a long time to accomplish manually. I needed some information about high school programs of study from across the United States. A simple request to ChatGPT generated a list of states that had the information I needed and I was then able to quickly verify manually that the results were accurate. This was a significant time saver for me and delivered data that I might have not been able to collect otherwise. 

I use artificial intelligence, principally Copilot, to provide a broad overview of an area of the law, in response to a crafted prompt, and then use its response to direct me to the specific law or court decisions which supports its response, and follow-up as necessary.  In doing so, I always verify the accuracy of the response prior to relying on it.  I also use artificial intelligence for document summary and as a drafting aid.

I used Copilot to help me create training documents for a recent electronic medical record (EMR) upgrade. Copilot helped me create step by step instructions for new features and enhancements in our EMR. I also asked for videos and found some. So some of my written procedures include step-by-step instructions, screenshots and a link to view a video.  It also provided training manuals from other clinics, I was able to use and not "reinvent the wheel". Timesaving!