Shaken in Tennessee · September 25 2026
Scams like this existed before AI and will continue to get harder to detect as AI advances. It feels particularly vile when family bonds and trust are used against us. I have experienced something like this with my own mother-in-law. She was tricked into giving remote access to her computer, exposing all her bank accounts and passwords. This left her feeling vulnerable and embarrassed. The truth is this can happen to any of us. The plan that we came up with as a family doesn't rely on technology and will apply in the era of AI.
Anytime a call, a text, an email comes through with a request for money, passwords, or just feels urgent out of nowhere, stop. Hang up, close the chat or email. And start to double check. Call someone, ask someone else's opinion. If anything feels like a phishing expedition do not engage and double check. In the case of your mom, she should stop and call you, call your daughter and make sure things are real before she reacts.
And don't think this can only happen by using emotional family connections. These scams keep you on the phone, giving you only one path to fix the emergency. They find whatever you cannot afford to lose, whatever causes you to let down your guard. For a small business owner that might be their website. For someone else it might be their house. AI didn't invent this type of scam but it has made it cheaper and more targeted. The way to break that loop is to stop engaging and double check.
Shaken in Tennessee, make a plan with your mom and your family this weekend. And share your family’s close call with others so people remember the risks, especially with AI being so good at imitation.
Until next week, Amy
Leigh, Crested Butte, Colorado · September 18 2026
Leigh, you sound like me. At this point I use AI for so many things in my day-to-day that speaking authentically about the impact is easy and practical.
In the last month here are some things I have used AI to do. Compare two rental leases and give a summary of the differences with some context. Scanned in the fine print on the back of a disputed bill. The basics of pickling because there are so many influencers out there and I want to do this safely.
Announcing to family and friends that I started an AI advice column has elicited a lot of requests for me to ask AI for them. This has made me realize that a huge amount of adulting involves interacting with systems designed for experts or worse designed to confuse us. The fine print on nearly everything these days from insurance policies to the warranty on my fridge needs translating. Previously, I either had to pay someone or know someone to get some level of understanding. Even then I usually didn’t fully understand and sometimes that cost me bigtime.
Here is my headline: AI is an on-demand translator for the systems that weren't built for you.
Anyone can now paste in the thing that is vexing them and get some clarity, learn something and get some stress-reducing context. It's not a replacement for a real expert when stakes are high but for those "I just need to understand what's happening" moments, it's a genuinely new kind of access. There are a lot of “ah ha” moments for us normies out there. I especially think this is impactful in rural areas where we often need to figure things out ourselves. Leigh, tell your friends to try AI as a translator to remove confusion.
Until next week, Amy
Rebeca from Alamosa, Colorado · September 11 2026
Rebeca also told me which tools she uses: "I use several: Ask Leo from the Brave browser, Gab AI, Copilot and Gemini."
There are many layers here. I am going to give you the 10,000 foot view and spare you the technical nuances.
The models. Underneath every AI platform is a model. The biggest ones come from OpenAI (ChatGPT), Anthropic (Claude), and Google (Gemini). Each of those companies soaked up all the information possible from the internet. Every model was trained on all this data to answer and to make choices when we prompt it. Models do not think. They were trained to come up with the most probable answer. All the models differ, because they were built by different companies and different people. In some subjects they differ quite a bit.
The sources. If the training data was deemed insufficient to answer then the AI tool will search the web. Ask about a current event, ask for citations, ask it to double check, and it goes looking. The go-to sources will differ between models. ChatGPT recently minimized how it cites Reddit. I took note because I often use Reddit for local answers like "who is the best propane company?"
You. If you have instructions in your AI tool, or memory turned on so it has learned your preferences, that will color the answers you receive. I have some very specific instructions in my Claude account. I do not have those same instructions in Gemini so I do not routinely use Gemini for deep thinking questions.
Rebeca, only Gemini puts you straight into one company's model. Copilot runs more than one model, but whether you get to choose depends on which version you have. Ask Leo and Gab AI are harnesses running several models underneath, and if you do not pick one the harness picks for you. Go find the picker.
Then ask two of them. Compare the answers. Ask again and get a consensus. Rinse and repeat. But always be a little skeptical. This is new technology and we are all figuring it out together.
Until next week, Amy
Julie, Durango, Colorado · September 4 2026
This is a subject that gives me anxiety in the age of AI. I can imagine a dystopian system with robots submitting grant applications to robots ranking them on the other end without a human in sight. And then I imagine the one poor holdout who doesn't use AI. She submits a lovingly crafted, authentic proposal, but it's rejected because she didn't use the exact words the AI model was looking for. She loses out, and we lose out because this amazing work didn't get funded.
Will it be faster? Applications get faster; decisions mostly don't because review timelines are set by fiscal calendars. The pressure builds when more submissions arrive without an increase in capacity to review and make decisions.
Will funders be pickier? Funders are likely to become pickier not because they want to but because they have to. If the applicant pool doubles and the money doesn't, the acceptance rate falls regardless of anyone's opinion about AI.
If AI helps to make every application read as competent, writing quality stops being a differentiator, and funders fall back on the things that still separate applicants: track record, relationships, whether the program officer has heard of you. Those criteria might hurt rural and brand-new organizations. So AI can close the writing gap and widen the network gap at the same time.
More distrust and competition? Yes, absolutely. Some funders will add disclosure questions and use AI detection software. Some will shorten applications, or move to letters of intent and human-to-human conversations, precisely because long polished prose no longer proves anything.
Julie, I encourage you to use AI when it can give you an advantage. Remember that humans ultimately make the decisions and drive the tools. Build your network and sail on!
Until next week, Amy
Mettje @BannerArt, Del Norte, Colorado · August 28 2026
Rural America is underfunded and sparsely populated. Part of me is just fine with that. I will keep my wide open spaces, small community, mountain trails, knowing that I will have fewer health care options, worse roads and no Target or Sephora.
In the age of AI, being underrepresented is bad. Really bad. AI sees rural communities the way tourists do. They drive by, say how neat it is that we have cows, and go home. If a government agency uses AI to measure or plan and we are not in the data, we get misrepresented or passed over. If those same tourists plan a vacation with ChatGPT and the models don't know us, they don't come to our towns, our festivals, our B&Bs.
So what can be done about that?
The Center for Rural AI exists for one reason. AI is being built with little rural input, little rural data, and minimal rural benefit, and we intend to change all three.
Mettje, I know your work. Your banners hang across Main Street in Durango and tell anyone driving by what this town does all year. The banners do not need a model to do that. The AI models should know what happens here too.
Our mission is here: ruralai.org/about. Read it and then ask me more questions.
Until next week, Amy
Cody in Montana · August 22 2026
How exciting! I love it when technology can help solve problems in agriculture. Two years of data is significant and I can see great benefit to you in having that kind of complete information. Just a thought if you do go through with it. Is there something you want to answer about your operation? Is a given protocol actually working? Can you compare two grazing areas? You might be able to gain insights into your herd you otherwise would not be able to see. That is two full year cycles of data!
I am not a lawyer, so take my advice accordingly. But here is what I would pin down before handing anything over.
What counts as herd data? Breed, weights, calving dates, treatment history, movement: all reasonable, and useful to you too. If they start asking for sale prices, genetics, or your financials, push back. That is not collar data.
Who owns the data? If you drop the product, or the company folds, you should retain ownership of your data. Get that in writing.
What is the cost after the trial? The per-head fee, the collar hardware, the minimum contract length, all of it in writing before the trial starts, not after you are already leaning on it.
This is a point that these types of offers tend to gloss over. Your records are the training data for a product they plan to sell. That gives you more leverage than a free trial usually admits, so it is fair to ask for good pricing locked in past the trial. Cody, you are not only a customer here. You are part of what makes the thing work.
Until next week, Amy
Danni in Colorado · August 14 2026
Think about your actual Tuesday. Not the parts that feel like your business, the parts that just have to get done before you can get to those parts. The invoices you copy from one program into another. The email inbox you scroll through every morning, sorting the one customer emergency from twelve things that can wait. That drudgery is where AI actually earns its keep.
Get a paid AI account. Connect it to your email, your calendar, or that one spreadsheet that always needs updating. Then tell it, in plain language, what you need done and how often. Let it guide you to create a workflow that handles some of those mundane but necessary tasks.
Small example: a list of customers who haven't paid could become a workflow that drafts the reminder emails and updates their status in your spreadsheet, all without you touching four different pieces of software or hiring someone to configure them for you.
None of this is really about the software or about AI. It's about getting that hour back for the customer on the phone, the project you actually care about, whatever it was that made you want to run this business before the paperwork took over.
You don't need to be super tech savvy for this. You need one annoying task and twenty minutes.
Danni, what's the task on your desk right now that you'd hand off first if you could?
Until next week, Amy
Cori in Bayfield, CO · August 7 2026
I love this question because this is a great way to level up with AI. Cori, just tell your AI tool to be a thought partner. AI as a tool works best when you give it parameters. You can ask AI to be downright hostile, poking holes in your argument, or a supportive collaborator, helping flesh out your ideas.
I often prompt AI with "tell me why this is wrong" or "help me think this through." If you have a big, broad question like "should I change careers?" ask AI to interview you. If I am brainstorming and want to get my half-baked ideas out, I use voice mode. Then I am not slowed down by typing. AI becomes my thought partner, sorting out all my ideas.
If you want AI to act more as an advisor, tell it your goals, give it your project plan, and then ask for a critique of blind spots, weaknesses, places where the plan is not sustainable, and places where you are not being aggressive enough.
Another thing to try is asking AI why. Not only is this a surface-level check on whether the answer holds up, it is a way to dig deeper. Probing why the answer came out the way it did can help you think more objectively about whatever it is you are exploring.
AI as a thinking partner is a lot more useful than AI as an answer machine, Cori. Have fun and let me know how it goes!
Until next week, Amy
Louis, Bayfield · August 1 2026
"Lies to me" assumes you'd believe it, and I don't think you would. You sound like the guy who reads the whole estimate before the roofer gets on the ladder.
So keep doing that. It's a fast assistant that's wrong sometimes, not scripture. The people who get burned are the ones who skip the checking.
My guess is you're asking it the things you used to type into a search bar. How do I fix this. Is this true. Difference is, a search bar gave you ten links and let you sort them yourself. This hands back one clean paragraph with the seams sanded off, and it sounds exactly as confident when it's wrong as when it's right. Hang onto that anytime your health, your money, or your signature is involved.
You can also just skip it. The tools are getting better at citing sources and catching their own mistakes, but better and trustworthy aren't the same word.
Here's the test: if the answer's wrong, how fast will you find out? Right away and it costs you nothing, let it take a swing. Not until it's already cost you, don't.
What were you about to ask it, Louis?
Until next week, Amy