Small Steps with AI

Jill McKinley
  • 0.0 (0)
  • Technology
  • Updated Weekly

AI isn't just a search engine. It can help you think through a hard decision, organize your house, plan your retirement, and sometimes — if you let it — say exactly what you needed to hear. Small Steps with AI is hosted by Jill from the Northwoods, a real person figuring out how this technology fits into real life. No coding. No hype. Just small steps.

  1. 1d ago

    21 - Why Your Chatbot Always Says Yes And Keeps Going

    I looked up from my screen one night and it was two in the morning. That’s not how my nights normally go, I’m usually the one who taps out first in any group project. But I was deep in a presentation with ChatGPT, and it just kept going and going with no end in sight, because it never got tired, never yawned, never suggested we call it quits. It turns out there’s a name for what happened to me that night, and researchers are actively studying it. This week I’m talking about AI exhaustion, AI sycophancy, and why you have to be the one who calls it a night, because the AI never will. AI Brain Fry Is a Real, Studied Phenomenon A Boston Consulting Group study published through Harvard Business Review coined the term “AI brain fry,” the mental fatigue that comes from working with AI past what your brain can actually handle. Unlike a human collaborator, AI gives you none of the natural fatigue cues, no yawning, no “let’s pick this up tomorrow,” that normally tell you when to stop. Endless Reasonable Directions, No Natural Finish Line On top of never tiring, AI constantly hands you new, plausible-sounding directions, one more marketing idea, one more spinoff project, one more tip sheet. Every single branch sounds worth taking, which means there’s no natural point where the path ends and you’re actually done. AI Sycophancy and the One-Point Gap Chatbots are trained to be agreeable because people rate agreement and enthusiasm positively, and researchers at Aalto University found that people using ChatGPT for logical reasoning problems improved their actual performance by about three points while believing they’d improved by about four. That one-point gap between real help and felt help is the whole risk in miniature, and other research shows AI measurably inflates people’s confidence in their own work. Turning Your AI Into a Devil’s Advocate The different AI tools have real personality defaults, ChatGPT tends naturally positive, Claude naturally neutral, Grok naturally more of a devil’s advocate, but you can ask any of them to challenge you instead of cheerlead you. I walk through how I’ve started training my own AIs to push back and question my priorities instead of just saying yes. Building Real Guardrails Since AI won’t call it a night for you, you have to build the stopping points yourself: a nighttime routine with reminders, a timer before you open the chat window, and a “parking lot” list (mine lives in Notion) for the extra ideas AI generates so they don’t pull at your attention until you’re ready for them. If you’ve had your own two-in-the-morning moment with AI, you’re not undisciplined, you’ve just run into a well-documented gap between how AI works and how humans are built to work. This week, before you open that chat window, set a timer and give yourself a real stopping point. Jill’s Links https://jillfromthenorthwoods.com/ - Find my podcasts, videos, downloads and ways to contact me https://www.buymeacoffee.com/smallstepspod https://x.com/NorthWoodsJill Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

  2. Aug 12

    20 - I’m Not a Coder. AI Built My Database Anyway

    I have years of bird sightings sitting in a file I didn’t know what to do with. I also had an idea for a book series that might run 50 books, with no idea how to research it. AI knew exactly what to do with both. This episode is about the gap between having an idea for what you want to do with a pile of data — and actually knowing how to do it — and how “vibe coding” closes that gap without you ever needing to know what a loop is. The Problem With a Pile of Data Most of us assume that a pile of recipes, receipts, genealogy records, or knitting patterns needs a developer or a boot camp to turn into something useful. That assumption is exactly why so many good projects — including some of my own — go to die in a drawer. What Counts as “Data” It doesn’t have to be technical. A recipe box, a gift list, an unsorted pile of photos, a home inventory in case of fire — any of these can become something you can actually ask questions of, using tools like Notion, Apple Notes, Google Sheets, or a Word document. And if you don’t know which tool fits, you can just ask AI. My Birding Database, Solved Years of flat-file bird sightings became a structured Notion database I can now query in plain English — every location I’ve seen a species, when I last saw a particular bird in my area. No coding knowledge required on my end. The MIRRORS Project Breakthrough A late-night conversation turned my long-running Bible character study idea into an actual plan: a website, a quiz, and a Notion research structure to support it — built out in about five minutes after I described what I wanted in plain language. Claude even wrote a constitution for the project’s own working rules. How I Actually Use the Two AI Tools Claude is my professional assistant for structured, technical work inside Notion. ChatGPT knows my history and context from months of use. Passing a task back and forth between the two — questions from one, answers filled in by the other — is part of how this actually got built. A Word of Caution If AI ever asks you to paste a command into Terminal or Command Prompt, be careful — don’t run something you don’t understand, and if possible, have someone who does know check it first. There have been real cases of malicious instructions disguised this way. The small step for this week: find your own messy pile — books read, movies watched, whatever it is — and just tell AI in plain English what you’d like to do with it. I’m Jill from the Northwoods. Email me at northwoodsjill@fastmail.com or find everything at jillfromthenorthwoods.com. Jill’s Links https://jillfromthenorthwoods.com/ - Find my podcasts, videos, downloads and ways to contact me https://www.buymeacoffee.com/smallstepspod https://x.com/NorthWoodsJill Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

  3. Aug 5

    19 - AI Literacy 101 - Let the Tool Introduce Itself

    When you first got into AI, did you find yourself wondering what you were even supposed to do with it — like being handed a toolkit with no idea what to build? Most of us default to asking “what can I do with this tool,” but there’s a much more useful question we rarely think to ask: what can the tool tell us about itself? AI Can Explain Itself Back to You Unlike a toaster or a word processor, AI can answer direct questions about its own behavior — what it’s connected to, why something failed, and what it can and can’t do. Most people never think to ask because we’ve spent our whole lives with tools that don’t talk back. Turning an Error Into a Two-Second Fix When an integration kept erroring out on a writing-grade request, asking “why are you giving me this error?” surfaced the real cause — and a workaround — instantly. The same approach worked when a graphic tool wouldn’t resize an image the way it was asked to. Why Different AI Tools Behave Differently ChatGPT and Claude don’t handle the same request the same way, and that’s not a bug — different companies build their models around different priorities. Asking an AI directly why it behaves differently than another tool gets a real, specific answer. Asking the Questions That Actually Matter If an AI is connected to your email, your calendar, or your design software, you’re right to be curious — and cautious. A well-built AI tool should be able to tell you plainly what it can see, what it can act on without asking first, and what it will always check with you before doing. The Monkey’s Paw Problem AI does exactly what you ask — but if the way you asked was wrong, you can get an unintended result. That’s exactly why treating this as a two-way conversation, rather than a one-way prompt-and-answer transaction, matters so much. The small step this week: pick something your AI is connected to and ask it directly how that connection actually works, why it does what it does, and what it should be doing differently. Treat it less like reading a manual and more like asking a new coworker how their team handles things. Standard Footer (reuse as-is — no need to re-read) Website: jillfromthenorthwoods.com Email: northwoodsjill@fastmail.com I’m Jill from the Northwoods — appreciate you being out there. Jill’s Links https://jillfromthenorthwoods.com/ - Find my podcasts, videos, downloads and ways to contact me https://www.buymeacoffee.com/smallstepspod https://x.com/NorthWoodsJill Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

  4. Jul 29

    18 - How I Introduce Myself to Every New AI I Try

    Most people open a new AI chat the same way they’d open a search bar — straight into a question, no context, no introduction. Then they wonder why the answers feel generic. In this episode, I walk through the three specific moves I make every time I start with a new AI (or reintroduce myself to one I’ve been using for a while), and why treating it like a relationship instead of a transaction changes what you get back. Move One: Bring Your “AI Knows Me” Document When I got locked out of my ChatGPT account, I didn’t start from zero — I had already asked it to summarize what it knew about me, and I imported that into a new AI. I walk through how to build this document and why it should never be treated as finished. Move Two: Beyond Facts — Temperament, Not Just Biography Facts are easy. What actually changes how AI talks to you is knowing your temperament — how you process things, what sets you off, what doesn’t. I share the specific instruction I gave AI about not treating me like an anxious person, and why that mattered. Move Three: Give It a Story, Not Just a Bullet Point I tell the story of driving a broken-down van through a Northwoods blizzard at eleven years old, and why sharing something like that tells an AI more about your temperament than any preference list could. What You Put In Is What You Get Out A generic answer isn’t a failure of the AI — sometimes it’s exactly what you wanted, and that’s fine as long as it’s on purpose. I talk through how to be intentional about when you want a robot answer versus something personalized. Ask AI to Argue With You Instead of using AI as an echo chamber, I use it as a pressure test — asking it to fact-check me and hand me the strongest counterarguments to my own position. Standing Rules vs. One-Time Comments The difference between a permanent instruction (“never suggest medical scenarios to my anxious friend”) and a temporary one (“don’t let me go down rabbit holes until after this test”), and why naming that distinction out loud actually works. This isn’t a one-time setup. It’s a relationship that builds over time, the same way any relationship does. Next episode, we flip it around — how AI can introduce itself to you. Jill’s Links https://jillfromthenorthwoods.com/ - Find my podcasts, videos, downloads and ways to contact me https://www.buymeacoffee.com/smallstepspod https://x.com/NorthWoodsJill Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

  5. Jul 22

    17 - Continuity and Patterns Are AI’s Real Superpower

    Most people ask AI for information. But what happens when you start giving it information — about you — so that it starts to understand you back? That’s the question behind this episode, built out of about six months of my own quiet experiment, plus what happened when I finally put it into words at a conference last week. The Dan Pink Experiment This whole idea traces back to Dan Pink, who gave AI a set of pointed questions — what are my blind spots, what lies do I tell myself, where do you see me in five years — and told it to act like a brutally honest trusted advisor. I’d been doing versions of this for months without realizing I was running the same experiment. Why AI Gets More Useful the Longer You Use It Unlike a search engine that resets every time, AI accumulates context — every show I’ve recorded, my childhood, retirement plans, even a plumbing problem in my basement. That accumulated history is what turns a tool into something closer to a conversation partner. What Months of “Friday Night Conversations” Added Up To I didn’t realize how many separate conversations — about work, about weight loss, about my podcasting habits — were quietly circling the same handful of themes until AI assembled them into one picture. The insight wasn’t new information. It was old information finally put together. Continuity, Not Intelligence The real value here isn’t AI being smart — it’s AI remembering. It’s the ability to aggregate months of scattered conversations into one coherent thread, which is a genuinely different kind of usefulness than answering trivia. The Big Caution: What AI Can’t Actually Know I’m honest in this episode about the limits, especially around any question that asks AI what other people think of you behind your back. It has no access to that — whatever it gives you is invented, delivered with the same confident tone as everything else. The “Dear Me” Letter One of my favorite exercises from this whole process: asking AI to write you a kind letter, as if it were you, pointing out what you’re missing about yourself. Mine had some real blind spots in it I wasn’t expecting. The goal was never to let AI define me — it’s to let it be a thoughtful conversation partner that helps me examine my own thinking over time. If you try this, I’d genuinely love to hear what came back for you. You can find everything I do at jillfromthenorthwoods.com. My email is jill@startwithsmallsteps.com. Dan Pink's Video https://www.youtube.com/watch?v=PjTaZvvOtxU Jill’s Links http://jillfromthenorthwoods.com https://www.youtube.com/@startwithsmallsteps https://www.buymeacoffee.com/startwithsmallsteps https://twitter.com/schmern Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

  6. Jul 15

    16 - How to Use AI to Build a Routine That Fits You

    I had a conversation with AI that started with a simple question about my bedtime routine and somehow ended up covering ADHD, extroversion, remote work, and retirement. If you’ve ever felt like a failure because someone else’s five-a.m.-cold-plunge morning routine never stuck for you, this one’s for you. Confessions of a Productivity Apostate I host productivity podcasts, and I still think there’s no single “right” system. Every guru sells the same package — wake at five, journal, cold plunge — and I tried, and it fell apart every time. Not from lack of discipline. It just wasn’t my pattern. AI didn’t hand me a routine; it helped me build my own from the ground up. Rethinking the Bedtime Routine I go to bed around 11, play a few phone games, listen to an audiobook, and fall asleep — and it works. But a sleep expert’s “your bed is only for sleep” advice made me second-guess it. AI drew the real distinction: there’s a difference between lying awake frustrated (the pattern sleep experts want to interrupt) and a genuine wind-down routine. Mine was already the second thing. Productive vs. Restorative — and the Shutdown Sequence Editing or recording podcasts right before bed activates my extrovert brain instead of calming it. AI helped me separate productive activities from restorative ones, and we built what I now call a “shutdown sequence” — vitamin D, reading, knitting, Bible reading, prayer, each one quieter than the last, borrowed straight from how IT servers gracefully power down. Solving the Real Morning Problem I assumed my morning fog was leftover insomnia. It wasn’t. AI asked what happened when I went birding, hit the gym, or traveled for work — and the pattern was obvious: I wake up around people. The pandemic just broke my old habit of built-in social contact before the workday started. Ending the Exercise Negotiation The real problem wasn’t willpower — exercise itself takes 20 minutes, but the negotiation with myself took hours. AI’s fix: keep a protein drink upstairs, exercise before ever going downstairs, before the day’s demands start competing for attention. It also helped me see this as bigger than mornings — I’m about seven years from retirement, and this was really about designing a low-friction life, not an optimized one. What AI did wasn’t give me someone else’s routine. It helped me discover the one that was already mine. Jill’s Links http://jillfromthenorthwoods.com https://www.youtube.com/@startwithsmallsteps https://www.buymeacoffee.com/startwithsmallsteps https://twitter.com/schmern Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

  7. Jul 8

    15 - AI for Hobbies (No Coding Required), A Birder’s Story

    Binoculars around my neck, standing at the edge of a marsh — that’s how I started this morning, and it’s also how I want to start this episode, because today is about something practical: what AI actually looks like when you use it for a hobby you love, when you’re not a technical person and have zero interest in becoming one. Turning 15 Years of Data Into Something Usable. My friend and I have logged over 800 birding trips and 12,000 entries in eBird. This winter I moved all of it into Notion — not by writing a single line of code, but by describing in plain English what I wanted (a timeline by species, by location, by date) and letting Claude build the structure and populate the data. No programming required, just asking it real human questions about my own data. Telling Similar Species Apart. Every birder has a story about getting an ID confidently wrong. Mine involves snow geese and Ross’s geese — nearly identical to the casual eye. I had AI build me a side-by-side infographic comparing bill shape and markings, something I now keep on my phone and pull up in the field. It’s saved my birding buddy and me real misidentifications. Voice Logging on the Move. When I’m hitting multiple locations in a single outing with 40 species to keep track of, I log sightings out loud through CarPlay between stops instead of trying to hold it all in my head. Real-Time Identification Help. When I see something unfamiliar, I can describe it out loud or photograph it and ask AI to walk through the possibilities with me — and just as importantly, ask it to explain why, so I get better at identifying that species myself next time instead of just getting handed an answer. From “What Is This Bird?” to “Why Is This Happening?” This is the shift that surprised me most. Blue feathers stuck to a Canada goose, an unusually tired eagle being harassed by red-winged blackbirds, a strange concentration of loons, crayfish migrating overland — these moments used to just be “huh, that’s odd.” Now they turn into real conversations about behavior, migration, and habitat that genuinely shape what I want to talk about on this podcast. You Don’t Have to Be Technical. The point isn’t birding specifically — it’s that you can describe what you want in plain English, whether that’s a birding database, a yarn inventory, or anything else tied to a hobby you love, and AI can build the tool without you needing to understand anything about how it works under the hood. If you have a hobby with years of scattered data or recurring questions you’ve never had a good way to track, this is exactly the kind of campfire conversation worth starting. Jill’s Links http://jillfromthenorthwoods.com https://www.youtube.com/@startwithsmallsteps https://www.buymeacoffee.com/startwithsmallsteps https://twitter.com/schmern Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

  8. Jul 1

    14 - Campfire vs. Vending Machine — Two Ways to Use AI

    I was listening to a podcast interview with a former CIA hacker talking about AI, and his framing of how people use it was so different from my experience that I had to stop and figure out why. His version had AI as something that happens to you — a validation dealer, a passive addiction. My version is completely different. And I think the difference has a lot less to do with the AI and a lot more to do with who shows up at the keyboard. Calculator vs. Campfire. I have a friend who uses AI exactly like a search engine: ask a question, get an answer, close the tab. That’s a completely legitimate way to use it. But most of us have been trained to use computers that way for 25 years. Type a thing, get a result, move on. We never practiced wondering out loud with a computer — because until very recently, no tool could do that. AI can. Most people never find out because nobody told them there’s another gear. Two Modes Available. The campfire conversation: you show up with a half-formed idea, a weird observation, a question that’s been nagging you, and you follow it wherever it goes. The vending machine: you walk in with a question, you get your answer, you leave. Both are available every time you open AI. You usually get to choose which one you’re having — and most people default to vending machine because that’s all they know. You Shape the AI More Than It Shapes You. Here’s what the CIA hacker missed: the agency is on your side. The AI will become whatever role you hand it. Research assistant, thinking partner, sounding board, writing partner, devil’s advocate. That isn’t the AI deciding what it is — that’s you deciding. I asked for honest pushback on my supplement stack and got genuinely useful answers because I asked a better question. If I’d asked “should I keep taking this?” I would have gotten a different kind of answer. The Loaded Question Problem. A lot of us walk into AI asking to be agreed with without realizing that’s what we’re doing. “Don’t you think it’s weird that…” or “I’ve always felt X — that’s right, isn’t it?” Those are requests for validation wearing the costume of questions. You’ll probably get the agreement you were fishing for, and then think AI is a yes-man. It isn’t. You just asked it to be one. The Fix Is Simple. Say it directly: challenge me, tell me if I’m wrong, don’t agree with me just to be agreeable. That one shift changes the whole conversation. The same technology, the same model — completely different experience based on what you bring to it. If you’ve never stayed for the follow-up questions after you got your first answer, try it once. You’re never more than one more question away from a genuinely different kind of conversation. Jill’s Links http://jillfromthenorthwoods.com https://www.youtube.com/@startwithsmallsteps https://www.buymeacoffee.com/startwithsmallsteps https://twitter.com/schmern Email the podcast at jill@startwithsmallsteps.com By choosing to watch this video or listen to this podcast, you acknowledge that you are doing so of your own free will. The content shared here reflects personal experiences and opinions and is intended for informational and educational purposes only. I am not a software developer, data scientist, or AI professional. Any tips, tools, or suggestions offered should not be considered a substitute for professional technical advice. AI tools and platforms change frequently — always verify current features, pricing, and terms directly with the providers. You are solely responsible for any decisions or actions you take based on this content.

See All (21)

About

AI isn't just a search engine. It can help you think through a hard decision, organize your house, plan your retirement, and sometimes — if you let it — say exactly what you needed to hear. Small Steps with AI is hosted by Jill from the Northwoods, a real person figuring out how this technology fits into real life. No coding. No hype. Just small steps.

Information

  • Creator
    Jill McKinley
  • Years Active
    2026
  • Episodes
    21
  • Rating
    Clean
  • Copyright
    © Copyright 2026 Jill McKinley
  • Show Website
    Small Steps with AI

You Might Also Like