AI Words in Plain English
You're in the right place. LookAgainAI's web address is magnifyai.madethis.app for now.
Sources checked Oct 29, 2026
You don't need to be an AI expert. But a few words come up again and again. Here's what they mean, and what to do about each.
- AI chatbot (or AI assistant): A program you type or talk to that writes back in everyday language. It predicts likely-sounding words. It doesn't look facts up unless it's connected to a search tool. What to do: Treat its answer as a first draft, not the final word.
- Prompt: The question or instructions you give an AI tool. What to do: Say what you want, who it's for, and ask it to tell you when it isn't sure.
- "Hallucination" (made-up answer): When AI states something false as if it were true. U.S. government guidance says this is a natural result of how these tools work, not a rare glitch. It can even invent sources. What to do: Check important facts, names, numbers, and links somewhere else.
- Agreeing too easily ("sycophancy"): AI tools tend to go along with you. Researchers found assistants often gave up a correct answer when the user simply pushed back. What to do: If an answer seems wrong, check it elsewhere or start a fresh chat. Don't argue.
- Trusting the machine too much ("automation bias"): People tend to accept a computer's suggestion even when it's wrong. In one study, experienced doctors did too. What to do: Slow down on anything that matters. A confident answer isn't proof.
- Hidden instructions ("prompt injection"): Text hidden in a web page, email, file, or picture that tries to steer an AI tool that reads it. What to do: If an AI summary suddenly urges you to click, pay, sign in, or share something, stop and look at the original yourself.
- AI agent (an AI helper that takes actions): An AI tool that can do things for you, like send messages, buy things, or change settings, not just answer. What to do: Let it look, not act. Approve anything that sends, pays, deletes, or shares.
- Deepfake: A fake photo, video, or recording made to look or sound like a real person or event. What to do: Don't rely on spotting glitches. Check through a way you already trust.
- Voice cloning: Using AI to imitate someone's voice. A scammer may use a short audio clip from public online content to imitate a voice. What to do: If a "family emergency" call asks for money, hang up and call the person back using a number you already know. If you believe someone is in immediate danger, call emergency services.
Try it now (about 5 minutes)
Copy the words in the box. Paste them into any free AI chat tool. Fill in the [brackets].
Explain the term "[word or phrase]" in plain everyday English, in 3 short sentences. Give me one everyday example.
Leave out names, account numbers and passwords. Check anything important somewhere else.
Sources (checked Oct 29, 2026):
- NIST AI 600-1
- Sharma et al., ICLR 2024
- RSNA, "AI Bias May Impair Radiologist Accuracy on Mammogram," May 2023
- OWASP, LLM01:2025 Prompt Injection
- OWASP, LLM06:2025 Excessive Agency
- NSA/FBI/CISA, Contextualizing Deepfake Threats to Organizations, Sept 2023
- FTC, "Scammers use AI to enhance their family emergency schemes," Mar 20, 2023
- FBI IC3, PSA241203, Dec 3, 2024