PLO 3 - AI Proposal
After attending Josh Cavalier's AI in L&D Intensive, I wanted to find a way for my team to integrate AI into our development workflows. One of the challenges was our company's current [stringent] AI policy, so I guided AI to support my research toward finding the best platform to meet our security needs to back up my proposal to leadership. This work aligns with PLO3: Apply research methodologies to answer questions and enhance practice in educational technology using accepted professional and institutional guidelines and procedures.
One of the major roadblocks to utilizing AI for our development work is security (hence our iron-clad AI policy) because we work primarily on government contracts, we can’t afford to have any proprietary information feeding into the large language model. So I used AI (ChatGPT and Gemini) to help me think about this problem in terms of research. I chose ChatGPT and Gemini for their ease of use and because I had pro accounts with both of them so I can get the AI to do more things for me than the free accounts could. I also chose to use both platforms to hopefully bypass any bias that may be inherent in the results of using one platform alone - for example if ChatGPT recommended itself over another platform, that could be a demonstration of bias).
​
I first used AI to think strategically about the research project and frame the situation as a research question. I then used AI to help me create a rubric to measure the platforms I intended to compare. I had AI run a search in "Agent Mode" to discover the top five platforms to compare. Then I told it run the results through the rubric and create a report. I took the findings and created a presentation for leadership with the goal of recommending a tool the training team can use to integrate generative AI into our development workflow. Both ChatGPT and Gemini recommended Copilot (which scored the highest on the rubric). While Copilot is not my favorite, and does not do as many cool things as Gemini or ChatGPT, it will at least be a step toward integrating AI if we are allowed to use it in the way I envision. What we really need is the ability to upload content into the AI to discuss and draw from that content.