Knowing AI's limits and pitfalls
AI is impressive but fallible. Knowing its limits protects you from nasty surprises.
Hallucinations: confident errors
An AI can make up false information while sounding completely sure of itself. It might cite a non-existent book, a wrong statistic, or a code function that doesn't exist. That's what's called a hallucination. The rule is simple: never take an important claim at face value. Always verify — especially facts, numbers, dates, and sources.
Outdated data and bias
An AI is trained on data up to a certain point — it might know nothing about recent events. It also reflects the biases present in its training data, which can color its answers on social or cultural topics. Keep in mind that its responses are neither neutral nor up-to-date by default. For current events, go to direct sources.
Privacy: be careful what you share
Don't paste sensitive information into an AI: passwords, banking details, confidential documents, other people's personal data. Depending on the tool, your inputs may be stored or used. Treat every conversation as potentially non-private. Anonymize sensitive information before asking for help, and check the tool's privacy settings.
Apply it now
- Always verify facts, numbers, and sources
- Don't rely on AI for recent news
- Stay alert to bias in its answers
- Never share sensitive or confidential information
- Read the tool's privacy settings