Feedback Loop
The short answer · 11 words
In short
A feedback loop returns results from a system as new inputs.
Flow-alerta.cab (2026): Feedback Loop. https://flow-alerta.cab/en/glossar/feedback-loop. Editorially reviewed on: 10 October 2026.
How much detail would you like?
Practical
Feedback can enable learning and improvement, but it can also increase distortions or instability.
Hypothetical example: Recommendations influence click data; therefore, the team also evaluates non-shown alternatives and long-term consequences.
SMEs should measure how automation decisions change later data, users and processes.
At a glance
- Core
- Feedback can stabilize or amplify. In AI systems, decisions often influence later data and thus future models.
- Use
- Feedback can enable learning and improvement, but it can also increase distortions or instability.
- Example
- Hypothetical example: Recommendations influence click data; therefore, the team also evaluates non-shown alternatives and long-term consequences.
- Limits
- View observed data as independent; self-amplifying biases; delayed effects; and ignoring missing control groups.
In context
Questions about this term
Is feedback always positive?
No; negatively stabilized frequently, positively reinforced.
Is user feedback undistorted?
Rarely completely.
All details and background
Feedback can stabilize or amplify. In AI systems, decisions often influence later data and thus future models.
ELI5 – Explained simply
The result changes the next round of the system.
Why it matters in practice
Feedback can enable learning and improvement, but it can also increase distortions or instability.
Example
Hypothetical example: Recommendations influence click data; therefore, the team also evaluates non-shown alternatives and long-term consequences.
Pitfalls and limits
- View observed data as independent; self-amplifying biases; delayed effects; and ignoring missing control groups.
Origin & history
In Good Data Analysis, Google describes how replayed click metrics can falsify the rating if the system itself influences these metrics.
Use in mid-sized companies
- SMEs should measure how automation decisions change later data, users and processes.
Related terms
- Control loop
- Monitoring
- Reinforcement learning
- Data bias
- Human-in-the-loop
Matching Flow-alerta service
AI opportunity analysis & roadmapAI consulting: Examine the effects of automated decisions on later data and key figures.Matching tool / use case
View use casesUse Case: Monitoring self-reinforcing effects in recommendation or prioritizationSources
Quellenprüfung: 10.10.2026. Definition und dokumentierte Einordnung abgeglichen; Beispiele beschreiben mögliche Anwendungen.
Editorially reviewed on: