Lesson 28 of 28 · Using AI well

Is AI safe to use?

AI is generally safe for everyday tasks if you treat it as a capable but fallible assistant. The main risks are confident wrong answers, private information you share being stored, biased output, over-reliance that weakens your own skills, and scams that use AI-made voices, images or messages. Check important facts, share less, and verify unexpected requests.

Wrong answers that sound right

AI assistants generate the most likely-sounding text, not checked facts. They can state wrong figures, misquote laws or invent sources with complete confidence, a problem called hallucination.

Use AI for drafts, ideas and explanations, and verify anything you will rely on, especially health, legal, financial or safety information. If an assistant cites a source, open it and confirm that it exists and says what the assistant claims.

Privacy: what happens to what you type

Anything you type or upload may be stored by the provider, reviewed for safety or used to improve future models, depending on the product, your plan and your settings. Treat a chatbot like an email to a company you do not know well.

Never paste passwords, ID or bank numbers, medical records or confidential work data into a personal AI account. Check the privacy settings, and at work use the tools your organisation has approved.

Bias and fairness

AI learns from human-made data, so it can repeat human biases: stereotypes in generated images, uneven accuracy across accents or languages, or skewed judgements about people. This matters most when AI informs decisions about jobs, loans, housing, education or health. Those decisions need human review and a way for people to challenge the outcome.

Scams, deepfakes and manipulation

AI makes it cheap to clone a voice from a short recording, fake a video or write a flawless phishing email. Deepfakes, realistic AI-generated audio, images or video of real people, are used in scams such as a "family member" calling in distress or a "boss" urgently requesting a payment.

Slow down whenever a message is urgent and involves money or secrets. Verify through a separate channel you already trust, such as calling back on a known number, and consider agreeing a code word with your family. Do not rely on AI detectors to settle the question; their results are often wrong.

Over-reliance and the bigger picture

Letting AI do all your thinking can weaken the skills you are trying to build, which is the main concern for students. Use AI as a tutor that explains and quizzes you, not as a ghostwriter, and follow your school's or employer's rules.

At the level of society, people also worry about misinformation, job disruption and how much power a few AI companies hold. Frameworks such as the US NIST AI Risk Management Framework (2023) and laws such as the EU AI Act set out how organisations should identify and manage these risks.

Example

Try it: make the assistant flag its weak spots

Try this prompt

Review your last answer. List every factual claim in it, mark each one as confident or uncertain, and tell me how I could check the uncertain ones using an independent source.

Use this after any answer you plan to rely on. The assistant's self-assessment can also be wrong, so treat its list as a checklist for your own verification, not as proof.

Key points

Treat AI as a capable but fallible assistant, and verify facts that matter.
Never share passwords, ID numbers, health records or confidential work data.
AI can reflect bias, so decisions about people need human review.
Verify urgent requests for money or secrecy through a separate, trusted channel.
Use AI to help you learn, not to skip learning.

Try it on Mindber

Lesson: what is an AI hallucination?

Glossary: AI acceptable use policy

AI governance, policy and risk

Find deepfake detection tools

Frequently asked questions

Is it safe to put personal information into an AI chatbot?

Share as little as possible. Depending on the service and your settings, conversations may be stored, reviewed by staff for safety or used to train models. Avoid passwords, ID and bank numbers, health details and anything confidential. If you must work with sensitive data, use a business plan with clear data protections or a tool your organisation has approved.

Why is AI bad for students?

AI is not bad for students in itself, but it is easy to misuse. If AI writes the essay or solves the problem, the student skips the practice that builds understanding, and may hand in errors or invented sources. Many schools set rules on AI use. Used as a tutor that explains, quizzes and gives feedback, AI can support learning instead.

Why do people say AI is bad for society?

Common concerns are misinformation and deepfakes, bias in automated decisions, privacy, job disruption, environmental cost and power concentrated in a few companies. Supporters point to benefits in productivity, accessibility, science and medicine. Many experts see both, which is why governments and standards bodies are building rules and frameworks to manage the risks while keeping the benefits.