Skip to main content
Tokenline

Line 1 · Station 6 of 6

Strengths & Limits of AI

Learn where AI shines, where it breaks, and why people stay responsible for how it's used.

  • 6 min
  • +50 XP
  • 2 knowledge checks

By the end of this station, you can:

  • List tasks where AI tends to be strong and where it tends to fail
  • Explain hallucinations, training cutoffs and brittleness in plain language
  • Explain why humans remain responsible for AI-assisted decisions

Powerful, but not magic

AI can do some things amazingly well and other things surprisingly badly — sometimes in the same conversation. Knowing the difference is what makes someone a smart AI user instead of a trusting one.

Where AI tends to shine vs. struggle

Often strong at

  • Finding patterns in huge amounts of data
  • Working fast and at scale, without getting tired
  • Drafting, summarizing and rephrasing text
  • Tasks that look a lot like its training data

Often weak at

  • Guaranteeing facts are accurate
  • Situations unlike its training data
  • Knowing about events after its training ended
  • Judgment calls involving values, fairness and context

Key terms

Hallucination
When an AI model produces information that sounds plausible but is false or made up.
Training cutoff
The point in time after which a model has no training data, so it may not know about more recent events.
Brittleness
When an AI system fails suddenly on inputs that are only a little different from what it was trained on.

Knowledge check 1

+10 XP on first try
You ask a chatbot who won a championship game played last weekend, and it confidently names a team. Why should you be cautious?

Myth vs. fact

  • MythIf an AI sounds confident, it's probably right.

    FactAI models can hallucinate — stating made-up facts, quotes or sources in a completely confident tone.

  • MythAI understands what it's saying the way people do.

    FactAI models process patterns in data. They can produce useful text without understanding meaning, having experiences or knowing whether something is true.

  • MythAI is objective because it's a computer.

    FactAI learns from human-created data and design choices, so it can reflect and repeat human biases.

  • MythOnce AI works in testing, it works everywhere.

    FactAI can be brittle. Performance can drop sharply when real-world input differs from its training data.

ExampleUsing AI's strengths wisely

You

Summarize my 6 pages of biology notes on cell division into a one-page review sheet.

AI assistant (sample)

The AI produces a neat, well-organized summary in seconds.

Why it matters: That's a great use of AI's speed. But before studying from it, compare it with your notes — a summary can drop a key step or mix up terms like mitosis and meiosis.

Knowledge check 2

+10 XP on first try
Which task is the best fit for an AI tool, with the least risk?

Key takeaways

  • AI is strong at speed, scale and pattern-matching.
  • It can hallucinate, miss recent events, reflect bias and break on unfamiliar input.
  • Confidence in AI output is not evidence of accuracy.
  • Humans stay responsible for decisions made with AI's help.

Sources for this lesson

  • National Institute of Standards and Technology (NIST). Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1, 2023. Full citation
  • University of Helsinki and MinnaLearn. Elements of AI. Full citation

Finish this station

Answer every knowledge check (right or wrong) to unlock completion.

  • Knowledge check 1 — not answered yet
  • Knowledge check 2 — not answered yet

0/2 checks answered