# kaal:claim:4855607-003

**Claim.** Deep learning models inadvertently learn and amplify whatever biases exist in their training data, so the composition of the training corpus, not the architecture, is the source of unfair or discriminatory outcomes.

**Type.** failure  **Support.** evidenced

**Holds when.**

- depends on the data used for training

**Source quote.**

> Depending on the data used for training, deep learning models can inadvertently learn and amplify biases present in the training data, potentially leading to unfair or discriminatory outcomes.

**From.** Wulf A. Kaal, *How AI Models are Optimized Through Web3 Governance* (2024), Model Overview: Deep Learning Models, page 9

**Cite as.** Wulf A. Kaal, How AI Models are Optimized Through Web3 Governance (2024). SSRN: https://ssrn.com/abstract=4855607

**Verify.** sha256 of source PDF `eb0b3e62374b45a8fa888c6bde9725e606bcb46cf4b5e74a6e851d9f25099113` at https://raw.githubusercontent.com/wulfkaal/Academic-Papers/main/papers/pdf/Kaal%20-%202024%20-%20How%20AI%20Models%20are%20Optimized%20Through%20Web3%20Governance.pdf

**Failure mode.** Training data bias amplification  (family: ai-model-and-training-failure)

**Topics.** ai-and-agents, education-and-practice

**Keywords.** deep-learning, bias-amplification, discrimination, training-data, fairness

**Related claims.**

- generalizes: https://wulfkaal.github.io/claims/5245185-028
- restated_by: https://wulfkaal.github.io/claims/4941807-015
- generalizes: https://wulfkaal.github.io/claims/5541658-013

**Canonical form.** This markdown file is the canonical hashed representation of the claim. Its sha256 is the content hash used for attestation.