Lesson 8: A closer look at Discernment | AI Fluency: Framework & Foundations Course

In this video, we'll dig deeper into the AI fluency competency of discernment. AI fluency means working with AI effectively, efficiently, ethically, and

safely. Discernment is specifically about evaluating AI outputs, processes, and behaviors. essentially your quality control system for AI collaboration. Discernment is your ability to critically evaluate what AI produces, how it produces it, and how it behaves. It's kind of the flip side of description. If description is about clearly communicating what you want, discernment is about deciding whether what you get back actually meets your needs. Developing your discernment helps you identify when AI outputs are valuable versus problematic. Recognize strengths and limitations and determine when outputs are ready to use or need more work. Doing this well requires both domain expertise, in other words, knowing enough to judge quality, and an understanding of how AI systems work, including their typical shortcomings. Remember, even the most advanced AI systems can make reasoning errors, produce factual mistakes, or act in unexpected ways. Your capacity for discernment acts as an essential safeguard. The most straightforward form of discernment is evaluating the quality of what the AI actually produces. When reviewing AI generated content, ask yourself questions like, is this factually accurate? Is it appropriate for my audience and purpose? Is it coherent and well structured? Does it meet my requirements? And does it add value or solve the problem I intended? We call this first concept product discernment, the ability to judge the accuracy and value of AI created output.

When interacting with AI, you need to assess not just what the AI produces, but how it got there. Some things to look out for include logical errors, lapses in the AI's attention, taking inappropriate steps, getting stuck on one small detail or interpretation and being unable to consider any alternatives, or getting trapped in circular reasoning. For example, imagine you're working with AI to expand on one of five outline options it offered for a document. After several rounds of back and forth ideiation together, you notice elements of rejected ideas being reinserted by AI. We call this kind of recognition process discernment, the ability to judge the quality and effectiveness of the AI's process. It's needed to ensure you and the AI are thinking in sync throughout the process, guiding the AI towards your vision for success. It becomes especially important for complex tasks where the correct answer isn't immediately obvious. In these situations, having trust in the process is everything. It's also often valuable to

evaluate and guide how the AI behaves during your interaction. We call this performance discernment. The difference between process and performance is maybe a bit subtle. You can think of process as the work the AI is doing, while performance is how well it is interacting with you while it does the work. When evaluating an AI assistant's performance, you can ask yourself, is there a better way for the AI to communicate with you for greater ease and productivity going forward? Is it providing the information you need in a helpful way? Does it respond well to your feedback and direction? And is the interaction efficient or unnecessarily complex? For example, is the AI asking too many questions when you need concise answers, or is it too brief when you need comprehensive information? We call this performance discernment, the ability to judge the quality of the human AI interaction, which helps you shape more productive working engagement with the AI systems. Of course

discernment doesn't end with evaluation. You also need to provide feedback in order to improve what an AI delivers going forward. When you identify problems in AI outputs, effective feedback includes specifying what the problem is, clearly explaining why it is a problem, providing concrete suggestions for improvement and revising your instructions or examples. In other words, when discernment flags a problem, better description is often the solution. But this isn't always the case. Sometimes you may need to rethink your delegation decisions because you might be using the wrong tool or approaching the problem in entirely the wrong way. To recap, product process and

performance discernment combine to form the discernment competency. Discernment works handinand with description. While description focuses on communicating your needs to AI, discernment evaluates how well those needs were met. Together, they form a continuous loop of instruction and evaluation that drives quality. By developing your discernment skills, you ensure that your AI collaboration remains guided by human judgment, a critical element of truly effective AI fluency.