Finding the next step…
piatra . institutepathwayspiatra . institutepathwaysFinding the next step…
A dataset’s collection, composition and intended use affect what conclusions a model evaluation can support.
Try the short introduction first, or bring equivalent reading experience. Obtain the specified text before beginning the longer reading.
Write a datasheet for a fictional dataset, including collection conditions, exclusions, consent questions and inappropriate uses.
Start here if the background is new. Equivalent experience is enough.
Introduces the question and vocabulary used in the longer selected reading.
Timnit Gebru · Jamie Morgenstern · Briana Vecchione · Jennifer Wortman Vaughan · Hanna Wallach · Hal Daumé III · Kate Crawford · paper
Abstract and Introduction, then the first developed argument or experiment.
Free full text
Open the readingDatasheets for Datasets. Research on dataset documentation, evaluation and harms in machine learning. Begin with how data was collected and which uses its documentation supports.
Answer this question in your own words: What should a dataset user know before training a model?
Name the claim, then explain why a reader might accept it.
A dataset’s collection, composition and intended use affect what conclusions a model evaluation can support.
What to look for
Common mistake: Summarising a biography instead of explaining the argument.
Start this pathway to keep your answers in My learning.
Timnit Gebru · Jamie Morgenstern · Briana Vecchione · Jennifer Wortman Vaughan · Hanna Wallach · Hal Daumé III · Kate Crawford · paper
Locate two passages about “What should a dataset user know before training a model” using the contents or index. Record their chapter/section names in your notes; this route assigns selections, not the whole work.
Free full text
Open the readingDatasheets for Datasets. Research on dataset documentation, evaluation and harms in machine learning. Begin with how data was collected and which uses its documentation supports.
What would make your activity a convincing example?
Write a datasheet for a fictional dataset, including collection conditions, exclusions, consent questions and inappropriate uses.
A useful response carries out the task: Write a datasheet for a fictional dataset, including collection conditions, exclusions, consent questions and inappropriate uses. It records the observation or passage used and explains how it bears on the question.
What to look for
Common mistake: Presenting an opinion without showing the example it rests on.
Start this pathway to keep your answers in My learning.
What is one conclusion this reading does not establish?
Test the boundary: Documentation improves scrutiny but does not by itself eliminate bias or authorise every reuse.
Documentation improves scrutiny but does not by itself eliminate bias or authorise every reuse.
What to look for
Common mistake: Treating a useful interpretation as an unrestricted fact.
Start this pathway to keep your answers in My learning.
Background, different viewpoints, and further reading.
Research on dataset documentation, evaluation and harms in machine learning. Begin with how data was collected and which uses its documentation supports.
Abstract and Introduction, then the first developed argument or experiment.