Responsible AI practices
Responsible AI Challenges in Traditional AI and Generative AI
Accuracy: bias and variance
Accuracy is the top problem developers face. A model only knows its training data, so poor training gives inaccurate results.
| Bias | Variance | Result | |
|---|---|---|---|
| Underfitting | High | Low | Too simple. Misses important features and performs poorly even on training data |
| Overfitting | Low | High | Memorizes noise. Great on training data, poor on new data |
| Balanced (the goal) | Low | Low | Captures the real patterns without the noise |
- Bias is the gap between the model's predictions and the true values.
- Variance is how much the model's predictions change with different training data, a measure of sensitivity to noise.
- The bias-variance trade-off is finding the balance between the two.
How to fix bias and variance errors
- Cross-validation: train on subsets of the data and test on the rest to detect overfitting.
- Increase data: add more samples.
- Regularization: penalize extreme weights to prevent overfitting.
- Simpler models: reduce overfitting. If a model underfits, it may be too simple.
- Dimension reduction (for example PCA): cut the number of features while keeping the information.
- Stop training early: so the model doesn't memorize the data.
Challenges specific to generative AI
- Toxicity: offensive or inappropriate content. Hard to define, and the line between filtering and censorship depends on context and culture.
- Hallucinations: plausible but false claims, such as invented scientific citations. They come from next-word sampling.
- Intellectual property: models reproducing training data verbatim, or imitating an artist's style.
- Plagiarism and cheating: essays and job applications written by AI. It's hard to verify who authored content.
- Disruption of the nature of work: worries that some professions will be replaced or transformed.