|
Hey there,
Welcome back. And a warm hello to anyone joining us for the first time.
AI knows more about you than you might think. Not in a scary way. In a "you have probably been typing things in without realizing what happens next" kind of way.
That is what this issue is about. Not to make you nervous about using these tools. You should absolutely keep using them. But a few simple habits will make you a much smarter user, and that is always worth five minutes of your time.
So what actually happens when you type something into AI?
When you send a message to ChatGPT, Claude, or any other AI tool, that conversation goes somewhere. Depending on the tool and the plan you are on, your conversation may be stored, reviewed by the company for safety purposes, or in some cases used to help train future versions of the model.
Most free versions of these tools have data retention turned on by default. That means your conversations are being kept, at least for a period of time. Paid plans often give you more control over this, including the ability to turn off training entirely or set conversations to delete automatically.
None of this is hidden. It is in the terms of service that, let us be honest, most of us click through without reading. But it is worth knowing.
The good news is that you do not have to stop using these tools or read a hundred pages of legal text. You just need to think a little differently about what you share.
Specific questions. Generalized details.
If you have been reading AI, Eh? for a while, you will know that being specific with AI gets you better answers. That is still true. But there is an important distinction worth making.
Being specific about what you need is great. Being specific about who you are is usually unnecessary. AI needs to understand the situation clearly. It does not need your name, your employer, your salary to the dollar, or your kid's school. Give it the context. Keep the personal details general. You will get the same quality of answer, and you will not have handed over information that has no business being in a conversation log.
Let's see what that looks like in real life
|
Sandra in Fredericton has a health question
Sandra has been having headaches and wants to know if she should see a doctor. Completely reasonable thing to ask AI about. Here is how most people approach it.
What most people type:
| "My name is Sandra and I'm a 47-year-old woman from Fredericton. I'm on metformin for Type 2 diabetes and I've been having really bad headaches on the left side of my head for the past two weeks, sometimes with blurry vision. Should I be worried? Do I need to see my doctor?" |
What Sandra should type instead:
| "I'm a woman in my late 40s with Type 2 diabetes managed by medication. I've been experiencing recurring headaches on one side of my head for about two weeks, occasionally with blurry vision. What might be causing this and when would it be advisable to see a doctor?" |
Same concern. Same question. Same quality of answer. No name, no city, no specific medication, no identifiable details.
One more thing worth saying here: AI is a useful starting point for health questions, but it is not a doctor. If something is worrying you, please reach out to your health care provider. No newsletter or AI tool should ever replace that conversation.
|
|
Dave in Regina wants a raise
Dave has a strong performance review and wants to build a case for a salary increase. AI is genuinely great for this. Here is where it goes sideways.
What most people type:
| "I work at Conexus Credit Union in Regina as a senior analyst and I've been there for six years. My current salary is $67,500 and I got a strong performance review last month. My manager is pretty tough and doesn't give raises easily. Can you help me build a case for asking for a raise to $75,000?" |
What Dave should type instead:
| "I work in financial services and have been in my current role for six years. I recently received a strong performance review. My manager tends to be conservative when it comes to salary increases. Can you help me build a case for asking for a raise, including how to frame my value and approach the conversation?" |
Dave still gets a well-crafted, useful response. His employer has no idea he is shopping for talking points. And his exact salary is not sitting in a conversation log somewhere.
|
|
Lena in Kelowna is worried about her son
Lena is going through a tough stretch with her teenager and wants some advice on how to handle it. This is exactly the kind of personal, emotional situation where people share more than they need to.
What most people type:
| "I'm really struggling with my 15-year-old son Tyler. He goes to Okanagan Mission Secondary and lately he's been withdrawing from the family, his grades have dropped and I found out he's been vaping. I don't know how to get through to him. What should I do?" |
What Lena should type instead:
| "I'm a parent struggling to connect with my teenage son. Over the past few months he's been withdrawing from the family, his grades have slipped, and I've recently discovered he's been vaping. I'm not sure how to approach the conversation without pushing him further away. Any advice?" |
Lena gets thoughtful, genuinely useful guidance. Her son's name, his school, and his personal situation stay private. A teenager can't consent to his struggles being shared with an AI tool. He deserves that consideration.
|
A quick word on Canadian privacy law
You may have heard the term PIPEDA. It stands for the Personal Information Protection and Electronic Documents Act, which is Canada's federal private sector privacy law. The short version is that it requires organizations doing business in Canada to be transparent about what personal information they collect, why they collect it, and how they protect it.
If you run a small business and you are using AI tools that handle customer information, that is worth paying attention to. You have a responsibility under Canadian law to handle that information carefully, and feeding customer details into a free AI tool without thinking it through could create real problems.
For everyday personal use, PIPEDA is less of a concern for you directly. But the habits we talked about above are still good practice. Being thoughtful about what you share is not just a privacy law thing. It is just smart.
The two-minute privacy check
Most AI tools give you some control over your data, but the settings are not always easy to find and they are rarely turned on by default. Here is what to look for the next time you have two minutes.
|
ChatGPT
Head to your account settings and look for a section called Data Controls. In there you will find options to turn off chat history, which also stops your conversations from being used to train the model. You can also find options to delete your conversation history entirely if you want a clean slate. The free version has some controls available, and paid plans generally give you more.
|
|
Claude
In your account settings, look for privacy controls related to conversation history and data usage. Claude's privacy approach is worth reading briefly as it is written in fairly plain language, which is not something you can say about every tech company's policy.
|
|
A general rule for any tool
Whatever AI tool you are using, look for words like Data Controls, Privacy Settings, Chat History, or Training in the settings menu. These menus move around as tools update, so the exact location may shift over time. But the options are usually there if you look for them. Spending two minutes finding those settings once is worth it.
|
One thing worth knowing before you turn everything off: if you disable chat history, the AI tool will not remember your previous conversations. Every session starts fresh. That is great for privacy, but it does mean you lose some of the continuity that makes these tools feel more useful over time. There is no perfect answer here. It is about finding the balance that works for you. Some people turn history off for sensitive topics and leave it on for everyday tasks. That is a completely reasonable approach.
Your challenge this week
Next time you open an AI tool, pause for just a second before you hit send. Ask yourself: does this prompt include my name, someone else's name, a specific employer, a dollar figure, or any detail that is more personal than it needs to be?
If it does, try this instead:
| "I am a [GENERAL DESCRIPTION — age range, role, situation] dealing with [THE SITUATION, without identifying details]. I am looking for [WHAT YOU ACTUALLY NEED]." |
Ready to use example:
| "I am a person in my early 50s dealing with a difficult conversation I need to have with a family member about money. I am looking for some advice on how to approach it calmly and constructively." |
Coming up next week
We are going to look at how AI can help you learn something new. Whether it is picking up a skill, understanding a topic that has always confused you, or just finally getting a straight answer on something you were too embarrassed to Google. Good one coming up.
|
AI assisted. Human led.
Whatever AI produces for you, always read it, verify it, and make it sound like you. AI is an incredible first draft machine. Your judgment, your voice, and your values are always the final layer.
|
As always, I read every reply personally. Have you ever caught yourself oversharing in an AI prompt? Hit reply and let me know. No judgment here. We have all done it.
If this one was useful, pass it along to someone who would get something out of it. The more Canadians who feel confident and informed about this stuff, the better.
Talk soon,
Chris
Founder, AI, Eh?
theaieh.ca 🍁
|
