Learning from people on the AI Learning track. People can rank outputs or correct labels. That feedback is data, with its own bias and delay. Write who is allowed to teach the model and how you stop one angry user from dominating.
This lesson assumes you already worked through Evaluation habits.
The idea in practice
Aggregate feedback. Down-weight duplicates. Do not train on a single thumbs-down in production the same hour.
A concrete check
goal = {
'track': 'AI Learning',
'lesson': 'Learning from people',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
Updating the model live from every click with no review. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
Write a rule for how many agreements you need before a correction enters training.
Self-check
- Say Learning from people in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.