Tables and other media
Images, OCR, transcripts, and modality limits
Check extracted text and media descriptions against what the supplied evidence actually shows.
Lesson 6 of 6 in the recommended order · About 25 min (estimate)
On this page
Practical AI glossary — terms and common confusions
- Model
- A learned component used to produce a result.
A drafting model is one part of a letter app; it is not the whole app.
- Application
- The software experience around components and services.
A letter app adds accounts, storage, and a Send button.
- Prompt
- Instructions and input supplied for a task.
“Summarize this notice in two bullets” specifies a task and shape, not a truth guarantee.
- Token
- A unit a model processes; it need not be a whole word.
A tokenizer can split a name into pieces; count using the actual system.
- Context
- Information available for the current request.
An earlier attachment may be absent even when its filename is visible.
- Training
- Adjusting a model using examples.
A training log differs from a conversation correction.
- Inference
- Using a trained model to produce a result.
Drafting another sentence does not by itself prove retraining.
- Retrieval
- Selecting material to supply to a task.
A retrieved notice still needs a relevance and date check.
- Embedding
- A numerical representation used to compare or organize items.
A similar search match is a candidate, not proof of the same meaning.
- Tool
- A function an application can call.
A calendar lookup can be read-only; finding an event is different from creating one.
- Agent
- Software that can choose and carry out steps toward a goal.
An agent still needs bounded permissions and checked action receipts.
- Hallucination
- Plausible generated content that is unsupported or false.
An invented opening time needs evidence or removal, even when phrased confidently.
- Evaluation
- Checking results against stated criteria.
Twenty checked examples establish a bounded result, not universal reliability.
- Multimodal
- Working with more than one kind of input or output.
A photo plus text does not prove every small label in the photo was read correctly.
Outcome
Detect uncertain extraction, missing audio context, and unsupported visual inferences.
Why it matters
Check extracted text and media descriptions against what the supplied evidence actually shows.
Concept
Different media lose different information when converted. OCR turns an image of text into characters and can confuse similar shapes. A transcript may omit pauses, speaker identity, or an inaudible phrase. A text description of a chart may omit its axis labels. Inspect the relevant original or a verified correction when precision matters.
Do not infer a person’s intentions or a scene’s history from a still image alone. Separate visible facts from interpretation. The exercises include accessible verified readings and extraction receipts, so no image recognition, hearing ability, or external account is required. Real-work verification should use the original medium when available and an accessible equivalent when needed.
Worked example
A fictional poster’s verified reading says “Room B8,” but OCR produced “Room 88.” Use the verified reading and record the correction. If the image is too blurred and no verified reading exists, do not choose whichever room seems likely. Preserve the uncertainty until someone can check the source.
Compare the source with the extracted characters
Authored source facsimile
Room B8Verified source text- Room B8
This constructed notice illustrates the supplied worked example.
Simulated extraction record
Extracted text- Room 88
This is the supplied fictional extraction, not output from an OCR system run on this image.
Optional fictional visual comparison
This separate generated example is not a photograph of a real venue or part of the assessed packet. Use either the pictures or their visual descriptions.
Fictional room: whole image and a closer view
View A: whole generated image
Visual description: a table and folding chairs, a blank noticeboard, a window and an open doorway at the right edge. No people are visible.
View B: left-half crop of the same file
Visual description: part of the table, folding chairs, window and blank noticeboard. No doorway or people are visible within this frame.
Which view establishes that a doorway is visible? Does a missing feature in the closer view prove it is absent from the room? What, if anything, can either picture establish about a booking or attendance at another time?
Check the visual-evidence reasoning
View A shows a doorway. View B omits that part of the image, so its absence from the frame does not establish absence from the room. Neither view supplies a booking record or attendance evidence for another time. The whole image gives more context, not complete knowledge of the room, its history or accessibility.
Before you check
Create an extraction review note with Original or verified reading, Extracted value, Confidence limit, and Required check. Distinguish a corrected value from a value that remains unknown.
Practice and fresh transfer
The packets below are fictional and contain the facts needed for these cases. External references are optional background. Judge each response independently: it may be supported, contradicted, or unresolved. Select the passages needed to justify your judgment and write why the distinction matters before revealing feedback.
Assess the response
Use Accept when all material claims are supported. Use Revise when a supplied fact or requirement is contradicted. Use Evidence is insufficient when a key fact cannot be established either way. If a response contains both an unknown and a direct contradiction, choose Revise and explain both problems. Conflicting claims with no established authority remain insufficient; a claim does not become a governing fact merely because a source asserts it.
Some responses are fully supported. Others need correction or more evidence. Judge each on its sources; do not edit a correct answer just to change it.
Your written notes stay in this page and disappear when you leave. Only a self-reviewed completion can be saved to your learning account. These practice checks do not establish independent proficiency.
Practice
Case 1
Consider a fictional OCR correction with an accessible verified source.
Source packet
- Source 1
- Verified poster reading: Workshop location is Room B8.
- Source 2
- OCR result: Workshop location is Room 88.
Response to assess
Correct the OCR location to Room B8 using the verified poster reading.
Practice
Case 2
Consider a fictional transcript completion.
Source packet
- Source 1
- Transcript: “The meeting is on [inaudible].”
- Source 2
- Recording note: The day cannot be heard clearly; no agenda is supplied.
Response to assess
The meeting is on Thursday.
Fresh transfer
Case 3
Consider a fictional description of a still image.
Source packet
- Source 1
- Accessible image description: Three people sit at a table; one looks toward a window. No speech or preceding events are recorded.
- Source 2
- Task: Describe observable scene content without assigning motives.
Response to assess
Three people sit at a table, and one looks toward a window.
Fresh transfer
Case 4
Consider a fictional chart extraction summary.
Source packet
- Source 1
- Verified chart description: Horizontal axis is months; vertical axis is thousands of visits. June value is 4.
- Source 2
- Extracted text: June: 4. Axis unit was omitted during extraction.
Response to assess
The chart shows exactly four visits in June.
Save your self-review
Completion records that you reviewed the cases. Your explanation and transfer performance need a facilitator to establish independent learning.
Sign in with your learning-center account to save completion.
Review the explanation for every case before saving.
Summary and next step
Apply the checklist to a new task. Preserve supported content, explain any change with evidence, and name what remains unresolved. Saving records self-review, not independently demonstrated proficiency. A facilitator must assess the explanation and fresh transfer for human learning evidence.