Free lesson · about 15 minutes
Using AI safely, in plain words.
This is the first session of the NeighborhoodOS program, written so you can do it on your own. No sign-up, nothing to install, and nothing you do here leaves your device.
A chatbot is a very big autocomplete.
You know how your phone suggests the next word when you text? Type “I’m on my” and it offers “way.” It learned that from patterns in what people usually type.
AI chatbots work on the same idea, at an enormous scale. They’ve read a huge amount of text, and they write an answer one word at a time by predicting what word usually comes next. They aren’t looking the answer up, and they don’t know whether it’s true.
Open a text message and tap the middle suggested word ten times in a row. The sentence sounds like you, but it doesn’t mean anything. That’s the machine working with no facts behind it.
It predicts what sounds right, not what is right.
It makes things up, and sounds sure when it does.
Because a chatbot writes whatever sounds likely, it will sometimes invent a law, a phone number, a date, or a quote that fits the pattern of a real answer. People call this a “hallucination.” The invented parts are written in the same confident voice as the true parts, so you can’t tell by reading.
That matters most when the stakes are high: a housing problem, a medical question, a bill, a letter to an official. If you cite a made-up code section in a letter to your landlord, your whole argument falls apart.
The Triple-Check Rule: anything specific (a number, a law, a deadline, a name, a link) gets checked at the original source before you rely on it. The city’s own website, the actual document, a real person at the office.
Ask a chatbot about the history of a specific building or corner in your neighborhood. Then check what it said against the library, a local historical society, or someone who’s lived there a long time.
What never to type into a free chatbot.
Free AI tools often keep what you type and may use it to improve their products. Treat anything you type as something a stranger might someday read.
- Social Security number
- Bank or card numbers
- Passwords or PINs
- Medical details about you or family
- Your lease, contracts, or legal papers
- Other people’s private information
If you need help with a private document, remove the names, addresses, and numbers first, or ask a person you trust.
The scams AI makes easier.
The voice clone. A few seconds of someone’s voice from a video can be enough to fake it. You get a call from what sounds like a grandchild in trouble who needs money right now.
What to do: hang up and call them back on the number you already have. Better yet, agree on a family safe word ahead of time, something a stranger couldn’t guess. Real family will know it.
The too-local message. AI lets scammers write messages that mention your street, your utility, or a local business, so they feel real.
What to do: don’t tap links in messages you weren’t expecting. Go to the official website yourself, or call the number printed on your bill.
Urgency plus money is the warning sign, however real it sounds.
Where it genuinely helps.
Treat a chatbot like a fast, overconfident helper: good for first drafts, never the final word.
- Good: turning a long, dense document into a shorter summary you then check.
- Good: drafting a polite letter or email you then edit.
- Good: brainstorming questions to ask at a meeting.
- Risky: anything about the law, your health, or money. Use it to find the right question, then ask a real source.
- Never: letting it decide something about a person, like who gets help or who to trust.
Three situations. What would you do?
Nobody sees your answers. They’re just for you.
Your pocket card.
Using AI safely
- It predicts what sounds right, not what is right.
- Check anything specific at the original source.
- Never type in ID numbers, bank info, passwords, health details, or legal papers.
- Urgent call asking for money? Hang up and call back on a number you know.
- Use it for drafts. You make the decisions.
neighborhoodos.org/learn