There are two kinds of AI conversations happening right now.

In one, AI is about to replace your job, your doctor, your kid’s English teacher and possibly your husband.

In the other, it’s a glorified autocomplete that confidently told someone to put glue on pizza.

Neither conversation is very useful if what you actually want to know is: can this thing help me get through a Tuesday?

Plenty of people are already trying to find out. In a Pew Research Center survey from early 2025, 34% of U.S. adults said they had used ChatGPT — roughly double the share in 2023. Twenty-six percent said they’d used it to learn something new. In Canada, Statistics Canada reported that the share of workers using generative AI on the job rose from 17% in September 2024 to 30% in July 2025.

So this isn’t a fringe hobby anymore. It’s a tool sitting on a lot of phones, often unopened, sometimes misused, occasionally brilliant.

Here’s where it earns its keep — and where it absolutely doesn’t.

First, a 30-second reality check on what these tools are

The AI assistants most people use — the chat-style tools, plus the versions now built into email, search engines, phones and word processors — are built on large language models. They generate text by predicting what’s likely to come next, based on enormous amounts of training data.

That makes them very good with language: rewording, summarizing, organizing, explaining, brainstorming.

It does not make them reliable sources of truth.

Even the companies building them say so. In a September 2025 research post, OpenAI described “hallucinations” as “plausible but false statements generated by language models,” and explained that standard training and testing tend to reward a confident guess over an honest “I don’t know.”

Read that again, because it’s the whole user manual in one line: these tools are rewarded for sounding sure.

Keep that in your back pocket for everything that follows.

Where AI is genuinely useful

1. The first draft of anything you’ve been avoiding

The email to your kid’s teacher. The polite-but-firm note to the contractor. The cover letter. The “I can’t make it, but thank you so much” reply you’ve been composing in your head for nine days.

This is the sweet spot. You give it the situation, the tone you want and the point you need to make. It gives you something to react to.

And reacting is so much easier than starting.

The trick: treat the output as raw material, not a finished product. AI-written prose has a recognizable sheen — slightly too enthusiastic, slightly too smooth. Cut the fluff, add one specific detail only you would know, and it sounds like you again.

2. Translating jargon into English

Your phone contract. A dense policy document from work. The paragraph in your home insurance renewal that seems important but is written like a hostage note.

Paste in the text (minus anything personal — more on that below) and ask: “Explain this in plain language. What does it require me to do? What should I ask about?”

This is one of the most practical everyday uses there is, because the source document is right there. You can check its explanation against the original line by line.

What it can’t do: tell you what the clause means legally for your specific situation. That’s still a job for a professional.

3. Life admin and the mental load

Brain-dumping is where these tools quietly shine.

Type out the chaotic list in your head — the dentist, the permission form, the birthday gift, the car registration, the thing for your mother — and ask it to sort everything by deadline, group it by errand, or turn it into a week-by-week plan.

It won’t do any of those things for you. But it can take the swirl out of your head and put it into a shape. If you’re the person in your house who carries the whole schedule, our piece on how to stop carrying the entire family calendar in your head pairs well with this.

4. Meal ideas from what you actually have

“I have half a rotisserie chicken, a bag of spinach, rice, a lemon and one sad onion. Give me three dinner options that take under 30 minutes.”

This works surprisingly well for brainstorming. Where you need to be careful: food safety, allergies and anything involving specific cooking temperatures or canning and preserving. For those, go to a public health or food safety source, not a chatbot.

5. Learning something, with you in the driver’s seat

AI can be a patient tutor. Ask it to explain a concept three different ways. Ask it to quiz you. Ask it to explain the thing your teenager’s math homework is asking, without just giving the answer.

The key is using it to test yourself, not just to read explanations. Asking “quiz me on this” is far more valuable than asking “summarize this.”

6. Rehearsing hard conversations

Asking for a raise. Setting a boundary with a family member. Returning something without a receipt.

You can ask an assistant to play the other person so you can practise your wording. It’s not a therapist, and it won’t predict how real humans will react. But saying the words out loud — even to software — can make the real version less shaky.

7. Planning, sorting and comparing

A packing list for a three-day trip. A rough itinerary. A pros-and-cons table for two options you’ve been going back and forth on.

Useful for structure. Not useful for facts you haven’t checked. If it tells you a museum is open Mondays, confirm on the museum’s own website before you show up.

Where AI falls down (sometimes spectacularly)

This is the part the breathless headlines skip.

It makes things up — convincingly

In 2024, a B.C. Supreme Court judge ordered a Vancouver lawyer to personally pay the other side’s costs for the extra work caused after she filed an application citing two legal cases that turned out not to exist. ChatGPT had generated them. The judge wrote that “generative AI is still no substitute for the professional expertise” lawyers need.

A trained professional, filing in an actual courtroom, got caught out. Anyone can.

Companies are on the hook for what their bots say — but that doesn’t help you in the moment

Also in 2024, B.C.’s Civil Resolution Tribunal ordered Air Canada to compensate a customer after the airline’s website chatbot gave him incorrect information about bereavement fares. The airline had argued the chatbot was “a separate legal entity that is responsible for its own actions.” The tribunal disagreed: “It should be obvious to Air Canada that it is responsible for all the information on its website.”

Good outcome. Still a hassle that went all the way to a tribunal.

The lesson for the rest of us: when a chatbot tells you a policy, a price, a deadline or a rule, find it in writing from the actual source before you act on it.

Health, money and legal questions need a human

An AI tool can help you prepare questions for your doctor, understand a term on a lab report, or organize your thoughts before meeting a financial adviser.

It should not be the thing deciding whether a symptom is serious, whether to stop a medication, how to file your taxes or what a contract obligates you to do. When Canada’s privacy regulators released their joint investigation into ChatGPT in May 2026, one of their findings was that users had been given insufficient notice that ChatGPT’s responses could be inaccurate — the regulators’ summary groups this under “factual inaccuracies and fabricated ‘hallucinations.'”

Anything you type may not stay between you and the screen

This deserves its own article, so we wrote one: Before You Paste Your Life Into a Chatbot: A Plain-English AI Privacy Guide. Short version: assume what you type could be stored, reviewed or used to improve the product unless you’ve confirmed otherwise in the settings.

How to check AI’s homework (without it becoming a second job)

You don’t need to fact-check a grocery list. You do need a system for anything that matters.

• Match the check to the stakes. Brainstorming dinner? Low stakes. A medication question, a legal deadline, a travel rule? Verify every time.
• Go to the original source. Government sites, the company’s own policy page, the actual document, a professional. Not another chatbot.
• Ask it for sources — then open them. Some tools can now search the web and cite links. Click through. A citation that doesn’t say what the AI claims it says is a red flag, not a footnote.
• Watch for suspicious confidence. Specific numbers, names, dates and quotes are exactly where fabrications hide.
• Ask “What might be wrong with this answer?” It won’t catch everything, but it sometimes flags its own weak spots.
• Give it the material. AI is more reliable summarizing a document you provide than answering from memory.

A quick guide to getting better answers

You don’t need “prompt engineering.” You need to talk to it like a capable new assistant who knows nothing about your life.

• Give context. “I’m writing to my daughter’s teacher about a missed assignment. I want to be friendly but clear that she was sick.”
• Say who it’s for and how long. “Three sentences, casual, for a group chat.”
• Ask for options. “Give me three versions: warm, neutral and firm.”
• Push back. “Too formal. Shorter. Less cheerful.” It doesn’t have feelings to hurt.

Let’s imagine a typical example: you ask for help with a birthday message for a colleague and get back something that sounds like a greeting card written by a committee. That’s not failure. That’s the first draft. Two rounds of “shorter, more like a real person” usually gets you there.

The Curious One’s take

AI is not magic and it’s not a menace. It’s a very fast, very articulate intern who never says “I’m not sure.”

Used that way, it can genuinely lighten the load: the drafts, the jargon, the lists, the endless small decisions that eat a day.

Used as an oracle, it will eventually embarrass you.

The women who get the most out of these tools aren’t the most technical. They’re the ones who stay in charge — who use AI to get started, then bring their own judgment to finish.

Which, frankly, is how most of us have been running things all along.

There’s more on how technology is reshaping everyday life across our Growth & Culture desk.

ASK MUDDY

What’s the one everyday task you’ve actually handed to AI — and did it help, or did you end up redoing it yourself?

JOIN THE MUDDY COMMUNITY

Life gets messy. We talk about it. Join the free Muddy community at muddystilettos.ca for new stories each week on the things women are actually reading, watching, learning and Googling.

ABOUT THE MUDDY GROWTH & CULTURE DESK

The Muddy Growth & Culture Desk covers books, shows, AI, culture and the internet rabbit holes worth falling into — and connects them to the bigger questions in women’s lives. We separate reporting from opinion and go back to the original sources.

Featuring The Curious One: books, Netflix, AI, culture and internet rabbit holes. Curious, reflective, sharp.

This article is educational and informational. It is not legal, medical or financial advice. For decisions in those areas, consult a qualified professional.

Sources

• “34% of U.S. adults have used ChatGPT, about double the share in 2023” — Pew Research Center (June 25, 2025)
• “Workplace artificial intelligence use: A profile of sociodemographic and job characteristics” — Statistics Canada, Insights on Canadian Society (June 17, 2026)
• “Why language models hallucinate” — OpenAI (September 5, 2025)
• Zhang v. Chen, 2024 BCSC 285 — Supreme Court of British Columbia
• Moffatt v. Air Canada, 2024 BCCRT 149 — B.C. Civil Resolution Tribunal (Feb. 2024)
• “Backgrounder: Summary of joint investigation into OpenAI’s ChatGPT” — Office of the Privacy Commissioner of Canada (May 6, 2026)
• “Glue on pizza? Google’s AI Overview shares wacky answers” — Silicon Republic (May 2024)