Definition
Can you now say how a machine sees, reads, listens, talks and moves?
This review brings together the key ideas of Domains of AI: how applications combine domains, predictive and generative AI, the three vision tasks and why vision fails, tokens and sentiment in language, speech recognition and synthesis, how voice assistants and chatbots work, and what makes a robot, followed by a scored quiz.
In simple words
Read the recap table, try the sorting activity, then take the quiz. Each answer comes with an explanation. If you miss one, go back to that topic and read it again.
Why it matters
Every domain does the same thing underneath: it turns pictures, words or sounds into numbers, finds patterns and gives an output. Seeing this shared idea makes any new AI product easier to understand.
How it works
| Idea | In one line |
|---|---|
| Domains in a chain | Apps pass results from vision to language to speech |
| Predictive and generative | One judges what exists; the other creates something new |
| Three vision tasks | Classification (one label), detection (boxes), segmentation (pixels) |
| Faces | Detection finds a face; recognition names a person and needs consent |
| Vision mistakes | New lighting, angles, rare cases and wrong clues |
| Tokens | Text is split into numbered pieces before a model reads it |
| Sentiment | Positive, negative or neutral; sarcasm is hard |
| Speech | Recognition: sound to text. Synthesis: text to sound |
| Voice assistant | Wake word, speech to text, understanding, action, reply |
| Chatbots | Rule-based: predictable. AI: flexible. Neither is a person |
| Robots | Sense, think, act, with sensors, a controller and actuators |
Try it yourself
Sort the tasks into domains
One last check: which domain does each task belong to?
Marking exactly which pixels are road
Turning a voice note into text
Splitting a sentence into tokens
Drawing boxes around cars in a photo
Reading a reply aloud
Deciding whether a review is positive
Carrying a shelf across a warehouse
Common mistakes
- Mixing up detection (where) with recognition (who).
- Thinking fluent speech or text means understanding.
- Thinking robot and AI mean the same thing.
Check yourself
Summary
- Vision, language, speech and robotics each turn one kind of input into useful output.
- Real products chain these domains together.
- Each domain has typical mistakes, so people check, decide and stay responsible.
Finished reading?
Sources: AI4K12: Five Big Ideas in AI · Elements of AI (University of Helsinki)