Five Ways AI Can Supercharge Your Productivity
We recently hosted Matt Stockton for an AI productivity retreat, here's what we learned.
I’ll admit, I’m not usually a first adopter. I’m usually a second-wave kind of gal. But I’m also not one to be left behind. I also multitask like my life depends on it. Because it does. So as the capability of AI tools has been increasing asymptotically this past year, I decided to spend some time over the Christmas break to take some courses and begin familiarizing myself with what’s out there. But in my quest to learn more, I thought, why not bring everyone I work with along with me on this AI journey. As a leader, I want my team to be equipped with the best tools to do the most creative work in the most productive way possible. So I called on Ann Arbor native and AI expert, Matt Stockton, founder of PragmaNexus, to lead an AI retreat for our faculty. We learned so much, that I asked Matt if I could share some of the tips we learned with all of you.
If you're anything like me, you've probably experimented with AI tools like ChatGPT or Claude. Perhaps you've asked for help drafting emails or summarizing research, and found the results intriguing but inconsistent. Perhaps the biggest take away for me is that the key to unlocking AI's potential is shifting from casual experimentation to purposeful adoption. So in that vein, here are five concrete ways AI can meaningfully enhance your productivity, distilled directly from practical insights Matt provided.
Treat AI Like Your Smartest (But Slightly Inexperienced) Research Assistant
Getting the most out of platforms like ChatGPT, Claude and Gemini means first thinking of AI not as an all-knowing oracle, but more like a highly capable graduate student who has read extensively but still needs clear, explicit instructions. When developing prompts, provide context, define your audience, set constraints (such as format and length), and offer examples. Doing so transforms vague results into focused, actionable content. One of the things we learned is that platforms like GPT generate most of their content from a “next-word prediction” algorithm, which in the past has led to a lot of highly annoying “hallucinations.” These are reasonable sounding but completely wrong bits information that can show up in responses. Fortunately, hallucinations have become less common as the models have improved, but they haven’t disappeared entirely. It’s important to understand that most of these platforms respond based on training data and predictive patterns, rather than live information. However, it can use web search when explicitly told to or when it recognizes that the question requires current or obscure information, which helps prevent against hallucinations. You can prompt this behavior by saying something like “please verify information with a web search and provide links.” It is important to specify web search when verifiability matters. Asking for live links helps you verify the accuracy of information yourself.
Use AI to Generate Structured Outputs
Avoid vague prompts like "summarize this paper." Instead, ask AI explicitly for structured outputs: tables comparing key studies, bullet-point summaries for meetings, or clearly formatted outlines for grant proposals. This small adjustment saves significant editing time and makes AI-generated content far more immediately useful. Tools like ChatGPT Projects enable you to maintain context and organize materials across sessions. You can ask for ranked information and hyperlinks to sources. For example, when analyzing research articles, request AI to generate a table listing each study's methodology, key results, sample size, and clinical significance. For meetings, you might ask AI to create a ranked list of priority action items with clearly defined deadlines and responsible parties. This structured approach significantly enhances clarity, speeds up decision-making, and ensures immediate applicability of generated insights.
Leverage AI’s Strength in Iterative Refinement
AI excels at revision. One helpful tip is to start a task by telling the platform what the task is and ask it to ask you questions about what information it needs to better accomplish your task. Once you have a rough draft, you can further refine with requests like "make this introduction more concise" or "shift the tone to professional yet approachable." You can also upload a sample of your past writing and ask the AI to match your tone and language. This is particularly helpful if you're trying to maintain consistency across a public-facing communication, such as newsletters, blog posts, or professional memos. I’ve even asked AI to mimic the rhythm and vocabulary of my own book chapters when drafting early versions of presentations—this works surprisingly well and can cut editing time in half. Once you’ve gotten a decent draft, ask the AI tool: “What would make this even better?” or “What would a professional editor do to improve this?” Often, the AI will offer insights you hadn’t thought of—like reorganizing sections, adding transitions, or eliminating fluff. Then, ask: “Revise this using your own suggestions.” It’s a quick way to elevate your output from solid to standout.
Use Cases and Specialized Tools You Should Know
While ChatGPT is a powerful generalist, there are other AI tools that can make specific tasks faster, easier, and more insightful. Here are some real-world use cases that go beyond casual experimentation and show how AI can truly improve your professional productivity. This list is tailored for academics, but still has some ideas for everyone.
Real time trouble shooting: Have a math problem you want to solve? Your computer giving you error messages? Try taking a photo of the problem and directly uploading it into your favorite platform and prompt the platform with your problem. I’ve used this to get step-by-step solutions for my son’s math problems as well as troubleshoot error messages on my MacBook.
Generating Custom Images: Need a conceptual diagram for a talk? Use AI image generators like DALL·E (in ChatGPT) for custom visuals. These tools can save hours otherwise spent digging through stock photo libraries. These tools have gotten significantly better in the past several months, and in particular their ability to incorporate text within the images to even generate full graphics for pamphlets or posters.
Writing Letters of Recommendation or Other Repetitive Writing that Follows a Format: This is one of the most tedious but high-stakes writing tasks we do. AI can help by drafting a professional, thoughtful letter based on a CV or set of bullet points you provide. I’ve used ChatGPT Projects to store templates and previous letters, so all I have to do is customize the voice and details for future letters. This is a big time saver.
Creating Presentations and Slide Decks: Prompt ChatGPT (or Claude) to generate a PowerPoint outline from a paper or idea. You can upload source documents to jump start the process. You can even ask it to format the content into specific slide types—like "Problem," "Solution," "Data," and "Call to Action." Then, just copy and paste into your slide software.
Create Unique Visual and Audio Content Utilizing NotebookLM: If you're a visual learner or simply trying to distill a dense set of documents, Google’s NotebookLM can generate mind maps, FAQs, and structured outlines based on your uploaded content. This is particularly helpful for planning lectures, writing manuscripts, or even preparing for patient education talks. For fun, also check out the NotebookLM audio overview feature which converts your uploaded content (PDFs, slides, documents, etc.) into a podcast-style audio summary featuring two different AI voices that discuss key points conversationally. We publish weekly internal “podcasts” for our team.
Deep Research with Search-Grounded Tools: For academics, tools like Consensus, ScholarAI, Elicit, and OpenEvidence are designed to surface peer-reviewed information directly from PubMed and other academic sources. Unlike ChatGPT, these tools are search-grounded and often include citations with links to original studies. Elicit excels at helping you identify and extract data points across multiple papers (e.g., sample size, intervention type), while OpenEvidence is especially useful for clinicians looking to make fast, evidence-informed decisions with real-time citations. Consensus and ScholarAI are specialized GPT tools available within ChatGPT. They're especially helpful when you're working on a literature review, writing clinical guidelines, or preparing background sections for a grant as they both access PubMed. Overall, these tools are in a different league when it comes to sourcing and providing citations. They’re purpose-built for surfacing peer-reviewed content, extracting structured data, and qualifying the strength of evidence — whereas ChatGPT’s general responses are often less consistent and don’t reliably provide transparent citations.
Leveling Up
The best way to learn how AI can help you is to simply start playing with it and create a culture of sharing use cases and best practices with people in your same field. The tools you are using today will likely be different even three to six months from now, so we need to embrace a continuous learning mindset. Matt provided some great resources for learning more:
3Blue1Brown: Large Language Models Explained (video)
A visually intuitive explanation of how LLMs function, great for understanding the basics of next-token prediction.
Andrej Karpathy: How I Use LLMs (video)
A practical walkthrough on using LLMs for tasks like summarizing PDFs, writing code and enhancing productivity.
ChatGPT Prompt Engineering Course (Andrew Ng & Isa Fulford)
A 90-minute course on practical techniques for improving AI outputs.
A curated collection of tutorials for applying AI tools.
Researchers, engineers, and thinkers discussing the newest in AI in real time.
Ethan Mollick’s Blog - One Useful Thing
Insightful writing on how AI is reshaping work, education and creativity.
The Upshot
If AI still feels like a novelty, it might be time to rethink your approach. These tools aren’t just for researchers or coders. They’re for anyone who juggles too much and needs a smarter way to work. Whether you're planning a vacation, redecorating a room, writing a speech, organizing your week, or summarizing dense material, AI can make tasks faster, clearer, and often more effective.
The real breakthrough happens when you shift from one-off tasks to integrated workflows: saving reusable prompts, organizing projects with memory, and letting tools support you like an extra brain. AI won't replace your judgment, but it will save you time. So don’t wait for the next big leap in AI. Start small, stay curious, and get comfortable experimenting. The future of productivity isn’t about doing more, it’s about spending your energy where it matters most.