Small Steps with AI

AI has a language of its own. This episode explains retrieval-augmented generation (RAG), APIs, tool calling, MCP, agents, prompt injection, human-in-the-loop oversight, and the larger questions of intelligence and consciousness, using everyday examples.

In this episode of Small Steps with AI, Jill from the Northwoods explores the topic through examples and practical observations.

• RAG and retrieving external information
• APIs, tool calling, and software access
• AI agents, risks, and oversight
• Intelligence, consciousness, and human judgment

[INSERT ORIGINAL SHOW-SPECIFIC DISCLAIMER HERE — exact approved wording not available in the uploaded transcripts]

Find more from Jill from the Northwoods at https://jillfromthenorthwoods.com/

AI Vocabulary Explained in Plain English
Small Steps with AI

LLMs, tokens, machine learning, agents, AGI: AI conversations can sound like a secret language. This episode explains common artificial intelligence terms in everyday English, with examples that make the technology easier to understand and use.

In this episode:
• What artificial intelligence means
• Machine learning and deep learning
• Generative AI
• Large language models
• Tokens and context
• Prompts and conversations
• Chatbots and AI assistants
• Agents and agentic AI
• AGI and staying informed

Explore more episodes at https://jillfromthenorthwoods.com

How AI Helped Me Adopt Two Cats

Can AI help with a personal decision without making the decision for you? After losing a beloved canary, I found myself weighing whether to welcome new pets into a suddenly quiet home. In this episode I share how an ongoing conversation with AI helped me sort through grief, timing, allergies, travel, cat-proofing, and whether to adopt one cat or a bonded pair. The result was two young brothers—and a useful example of AI as a thinking partner rather than a substitute for human judgment.

IN THIS EPISODE
• A different kind of AI story
• Grieving a canary and considering a new pet
• Why timing mattered
• AI as a campfire conversation
• One cat or two?
• Finding the right rescue cats
• Preparing the house and settling in
• What AI did—and did not—decide

CHAPTERS
00:00 Introduction: How AI Helped Me Adopt Two Cats
02:36 Grieving a canary and considering a new pet
05:11 Why timing mattered
07:49 AI as a campfire conversation
10:21 One cat or two?
12:55 Finding the right rescue cats
15:33 Preparing the house and settling in
18:07 What AI did—and did not—decide
20:45 Putting the ideas together
23:28 Practical lessons and reflections

REFERENCES AND RESOURCES
Tools and topics mentioned: ChatGPT; Petfinder; bonded cat adoption; pet safety and cat-proofing. AI can help organize questions, but adoption decisions and animal care remain human responsibilities.

Jill From the Northwoods
https://jillfromthenorthwoods.com

AI can explain, compare, draft, and troubleshoot, but useful assistance does not require handing over every private detail. I walk through a practical circle of trust for information, safer ways to share documents, and the extra care needed when AI suggests computer commands or advice with serious consequences.

Ask Three Questions Before Sharing
What does AI actually need to know? What happens if its answer is wrong? What would happen if the information were exposed? These questions help separate relevant context from unnecessary identifying details and keep our judgment involved as tools become more capable.

Red, Yellow, and Green Information
Passwords, login codes, recovery keys, and full account identifiers belong in the red category. Health, financial, employment, and other people's private information require careful judgment in the yellow category. Generalized examples and nonidentifying context can often provide a greener way to get useful help.

Protect Other People and Work Materials
A friend's diagnosis or a coworker's private situation is not ours to disclose casually. Workplace rules also matter: permission to use an approved company tool does not automatically cover a personal account. I share why checking the rules and removing unnecessary names can make a difference.

Look Beyond the Question in a Screenshot
AI may notice more than the detail we intended to show. Documents and photographs can reveal addresses, account numbers, background objects, and other information. Crop to the relevant area and check the finished image before sharing rather than assuming the rest will be ignored.

Keep Judgment over Actions and Advice
A command that fixes a problem can also delete files, change permissions, or modify a database. Understand it, make backups, and seek qualified help when needed. For medical, financial, and legal questions, AI can help organize information and prepare questions, while important decisions deserve professional verification.

Before your next upload or powerful command, pause and ask the three questions. Share only what is needed, understand proposed actions, and increase verification as the stakes rise.

[email protected] and jillfromthenorthwoods.com

A note on this podcast: I’m not a software developer or an AI professional, just someone who uses these tools every day and shares what’s working for me. Test anything I mention against your own situation before you rely on it.

TIMESTAMPS

00:00 Using AI includes knowing when to pause
01:59 Three questions for safer AI use
03:05 Relevant context without identifiers
06:54 Red information: credentials and codes
12:07 Yellow information and privacy
14:56 Other people and workplace rules
16:39 Green information and general examples
18:19 Crop documents and check screenshots
20:53 Understand commands before running them
23:42 High stakes and your final judgment

A listener asked me how AI might help someone with low vision, and that sent me into a series of experiments. I wanted to know what these tools could actually do with tiny print, household products, photographs, instructions, and ordinary objects, and just as importantly, where they could fail.

AI Can Help With More Than Reading
A listener question sent me looking at how AI can help people who are blind or have low vision. I tested ordinary situations where a camera, voice conversation, and follow-up questions can turn visual information into something more usable.

From Tiny Print to Useful Answers
Magnification and text recognition have been around for years, but generative AI adds another layer. It can read instructions, explain what they mean, answer questions about an object, and sometimes adapt the information to what you actually need.

The Tools Are Getting More Conversational
I look at Apple Magnifier and voice features, Microsoft Seeing AI, Be My Eyes, and smart glasses. The interesting change is not simply that technology can recognize something, but that you can keep asking questions about what it sees.

AI Can Be Right and Still Miss Something
One of my tests produced accurate information while skipping a line entirely. That is an important warning for visual accessibility because a person who cannot see the original may not know that anything was omitted.

Match the Verification to the Risk
A wrong shirt color is inconvenient. A wrong answer about medication, allergies, finances, legal papers, or physical safety can matter a great deal, so higher-risk situations need another source of verification.

Try One Ordinary Thing
AI does not replace established accessibility tools, training, specialists, screen readers, or human help. It can be one more flexible tool, and the best first experiment is something small and low stakes.

The useful shift is from a tool that simply magnifies or reads to one you can question. Start with something ordinary, keep the stakes low, ask follow-up questions, and learn where AI fits alongside the accessibility tools and people you already trust.

Contact: [email protected] | jillfromthenorthwoods.com

A note on this podcast: I'm not a software developer or an AI professional, just someone who uses these tools every day and shares what's working for me. Test anything I mention against your own situation before you rely on it.

TIMESTAMPS

00:00 AI and low vision
03:44 From magnification to conversation
05:32 Testing AI on everyday objects
07:35 Apple Magnifier and visual questions
09:42 When a correct answer is incomplete
11:31 Voice control and hands-free help
13:35 Seeing AI, Be My Eyes, and smart glasses
16:55 The confidence problem
19:34 Low stakes versus high stakes
23:24 Start with one ordinary task

Everyone wants to ask "which AI subscription should I buy?" I think that's the wrong first question. The right one is: what did the free version actually stop me from doing? This episode walks through how to figure that out before you hand over $20 a month to anyone. TRY FREE FIRST, ON PURPOSE Spend a week using a free AI tool on real tasks you'd normally do, gardening spreadsheets, birding lists, rough retirement math (no account numbers, just ballpark thinking). Notice exactly where you hit a wall. THE ONLY QUESTION THAT MATTERS Not "which AI should I subscribe to," but "what did free prevent me from doing?" If the answer at the end of the week is nothing, you've saved yourself a subscription. WHERE FREE TENDS TO BREAK DOWN Usage limits that cut you off mid-project, features locked behind a paywall like larger projects or deep research, or a specific integration that would save you real hours every week. Those are legitimate reasons to upgrade. "Someone on YouTube said so" is not. HOW THE MODELS DIFFER RIGHT NOW ChatGPT tends to win on personalization, image generation, and third-party integrations. Claude tends to win on writing quality but has tighter free-tier limits. Perplexity has a reputation for strong sourcing. Each has real free-tier limitations worth testing for yourself, since it changes constantly. A REAL EXAMPLE OF WASTED MONEY A friend bought a full year of an AI subscription and realized afterward she didn't write professional documents, didn't code, and wasn't using it for anything a free tier couldn't already do. WHEN IT BECOMES OBVIOUS YOU NEED PAID Running multiple podcasts made a paid account an easy call: grouped projects for thumbnail work, heavier research needs, and specific automations that save real time each week. The bottom line: don't subscribe because you're worried about missing out. Subscribe because you can name the exact wall you keep hitting. If you've got a story about deciding between free and paid AI, I'd love to hear it. Email me at [email protected], 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 [email protected]

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.

I consume a genuinely ridiculous amount of content, books, podcasts, videos, articles I've been saving for twenty plus years into a filing cabinet in my basement that I haven't opened in probably as long. For most of my life, the weak link in all of that consuming was always me: I'd underline something, screenshot a quote, tell myself I'd come back to it, and then watch it disappear into the enormous warehouse of things I once found interesting. This episode is about how AI changed that, not by writing content for me, but by helping me actually use everything I take in. The old model of consuming content was broken by design Books, podcasts, notes, saved articles, they all sat in a system (my filing cabinet, in my case) waiting to be remembered and retrieved, and most of it never was. I moved from Pocket to Notion a while back, and now every transcript and saved article lands in one searchable place instead of a drawer I forget exists. A real example, fact checking a nutrition article while mowing the lawn I ran a health article's advice past AI and asked which parts were actually backed by evidence versus which weren't. It agreed on some things (choline from eggs, for instance) and helped me push back on claims that didn't have real research behind them, including asking for sources when something felt shaky. A real example, turning career advice into something personal I listened to Mel Robbins interview Carla Harris on career advice by decade of life, then had AI pull out what actually applied to me at my age and stage, including some hard self reflection about growing up needing to please other people rather than asking myself what I actually wanted. Why this makes creators better instead of making everyone the same The fear that AI will homogenize content is real if you use it lazily, ask ten people to "write a podcast about productivity" and you get ten forgettable scripts. But the better use starts before the writing: what irritated you, what you noticed, what connected to something you've been thinking about for months. That's the part AI can't generate on its own, it can only help you dig it out. AI as a sparring partner, not a ghostwriter Because everything I save lives in one system now, AI can connect a career podcast to my own ikigai episode from months ago, warn me when a draft has an edge I don't actually want, or collapse a dozen saved stretching articles into one usable routine, all without me having to remember any of it existed. What I actually want from AI as a creator Not speed, and not friction free content. I want something that helps me notice more, question more, and remember what I already know, so that by the time I turn on a microphone, I have something that's actually mine to say. If you're a content hungry person who's tried and failed to build a "system" for everything you read and listen to, this one's for you. I'd love to hear how you're using AI to think, not just to produce, send me your version. 00:00 Intro, twenty years of saved articles I never look at 01:59 Why the old model of consuming content breaks down 03:28 Moving from Pocket to Notion, building a real system 04:27 What AI actually adds beyond summarizing 06:58 Case study, fact checking a nutrition article and a career podcast 11:53 Why bringing yourself to the process is what prevents generic content 16:21 Connecting new ideas to old ones, the ikigai example 21:19 Turning saved articles into one usable stretching routine 22:21 How AI helped me notice my own "harsh edge" while writing 26:22 Why not everything needs to become content, wrap up [email protected] 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 [email protected]

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.