Write your response in your own notes, then compare it with the sample answer. This practice is separate from the scored exam and does not affect your certificate.
Practice 1
Turn on your device's built-in screen reader and move through a simple webpage heading by heading without looking at the screen. Find one image on the page. Write down what the reader said about it, whether that was genuinely useful or just 'image', and who would need to fix it if it was unhelpful.
Compare with a sample answer
A good answer names the reader used (Narrator, VoiceOver or TalkBack) and says the page was moved through heading by heading. It records what was said about one image and judges whether that was useful or merely 'image'. If the description was unhelpful, it points out that a screen reader can only describe what the page tells it, so the usual fix is the page's missing labels and not the learner's settings.
The sample follows the exercise steps and the callout: navigation by headings, a judgement on the image description, and the usual cause of a poor reading being the page.
Practice 2
Dictate three or four ordinary sentences, such as your plans for the day. Read back the typed text without correcting it first, then list every wrong word. Say which kind of word caused the most mistakes, and how often you would read back a message that contains many names or numbers.
Compare with a sample answer
A good answer reads the text back before correcting anything, lists each wrong word, and notes whether names, numbers or less common words caused most of the errors. It says that for messages with many names or numbers the read-back should happen every time, not only for important documents, and that dictated numbers should be read back specifically before sending.
The sample follows the exercise order (read back first, then list errors, then note the kind of word) and its reflection that names and numbers need a careful read-back every time.
Practice 3
Play at least one minute of a recording you already know roughly, with live captions on. Record what kind of error appeared (a wrong word, a missed word or a misattributed speaker) and whether any name or technical term was captioned correctly. Say whether the errors would have mattered if you had relied on captions alone for a job interview or a medical appointment.
Compare with a sample answer
A good answer compares the captions with what the learner already knows was said, names the kind of error found, and checks any name or technical term. It then answers the honest test: if the errors would have mattered in an important situation, captions alone are not enough there, and a record that matters needs a proper transcription service.
The sample follows the exercise steps and its reflection, and adds the course's point that live captions are not a record.
Practice 4
Get an AI description of a photo you know well. Compare it with what you know is in the photo and list anything it got wrong, invented or missed. If the photo has a sign, label or handwriting, check whether the described text matches exactly.
Compare with a sample answer
A good answer lists specific details the description got right and any it got wrong, invented or missed. For a sign, label or handwriting it compares the described words and numbers with the real text and treats the AI's reading as a draft to verify, not a confirmed transcript.
The sample follows the exercise steps and its reflection that text read from a photo is a draft to verify, particularly where numbers are involved.
Practice 5
A relative sends you a photo of a notice on a machine at work, and an AI description says it gives a phone number and a closing time. Using the course's checklist, write what you would do before relying on it.
Compare with a sample answer
A good answer treats the phone number and closing time as unconfirmed and verifies them against the photo, because AI readings of text and numbers are a draft. It does not assume the description covered everything, gets a second confirmation because a safety or work notice is consequential, and asks again or tries a different tool if the description does not match what it expects.
The sample applies all four checklist items: verify text, do not assume completeness, get a second confirmation for consequential material, and ask again or try another tool on a mismatch.
Practice 6
Pick one of the four tools (screen reader, dictation, live captions or image description). Write a short plan: one situation where you would use it with no extra check, one where you would add a quick human check, and one where you would not rely on it alone.
Compare with a sample answer
A good plan keeps using the tool and matches the check to the stakes. For example, live captions can be used to follow a casual video with a clear speaker; a quick check is added for names and technical terms; and for a record that matters, such as a medical consultation, a proper transcription service is used instead of live captions.
The sample reflects the course's rule that the tools are useful but not infallible, and that the skill is knowing which situations need a human check and which do not.