Lesson 4 of 7 · about 4 minutes
What can AI do, and what can't it?
The big idea: AI is great at finding patterns in lots of information. But it does not know things the way you do, and it makes mistakes.
Things AI can be good at
Here are some jobs that AI tools are built to help with:
- Turning speech into text, and text into speech.
- Translating between languages.
- Spotting things in pictures, like animals or objects.
- Writing, summing up, and answering questions.
- Helping scientists study proteins, the tiny parts that help living things work.
Things to be careful about
AI also has limits. Keep these in mind:
- AI makes mistakes, and sometimes it makes things up.
- It may not know about things that happened after its training, unless it can search for new information.
- Sounding smart is not the same as being right.
- AI is not your family or friend. It does not know you the way they do.
- People, not AI, are responsible for important choices.
Why AI gets things wrong
AI learns from examples. Sometimes those examples are wrong, missing pieces, or unfair. Then the AI can learn the wrong patterns.
That is one reason people test AI carefully before trusting it.
Try this: Fact or fake?
- With a grown-up, ask a chatbot a question you already know the answer to. Try "How many legs does a spider have?"
- Check the answer in a book or a trusted website.
- Was the chatbot right? How could you tell?
Remember
- AI is good at patterns in words, pictures, sounds, and numbers.
- AI can be wrong and can make things up, so check important facts.
- People, not AI, are responsible for important choices.
Quick quiz
Pick an answer to see if you got it. Nothing is saved.
0 of 3 answered.
Quick quiz
Circle the best answer.
1. Which job is AI often good at?
- Turning speech into text
- Knowing exactly how you feel
- Never making a mistake
2. A chatbot tells you a fact for a school report. What should you do?
- Copy it right away
- Check it with a trusted source
- Ask the chatbot if it is sure, then copy it
3. Why might AI learn something unfair?
- Its examples were unfair or incomplete
- Computers are mean
- AI cannot learn anything
For grown-ups: sources and going further
The "good at" list matches kinds of tools in the catalog, such as speech recognition models and protein models used in research. The cautions come from the explainers on how language models work and how models use tools such as search.
- Why fluent answers can still be wrong
- How AI models use tools
- Speech, vision, and multimodal AI hub
- AI for science hub
Written for readers about 8 to 12, last reviewed 2026-10-11. How these lessons were made.
Answer key
- 1. A, Turning speech into text. Speech models turn spoken words into text. AI still makes mistakes, though.
- 2. B, Check it with a trusted source. Chatbots can be wrong. Check facts in a trusted book or website, or ask a teacher.
- 3. A, Its examples were unfair or incomplete. AI learns from its examples. Unfair or incomplete examples can teach it unfair patterns.
https://unitedstatesofamericasuperintelligence.com/kids/what-ai-can-and-cannot-do/ · Source: USASI catalog (unitedstatesofamericasuperintelligence.com), CC BY 4.0. Independent project, not a U.S. government website.