Epistemic Artificial Intelligence is Essential for Machine Learning Models to Truly 'Know When They Do Not Know'
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:RP7CZJHArecord.jsonopen to challenge →
read the original abstract
Despite AI's impressive achievements, including recent advances in generative and large language models, there remains a significant gap in the ability of AI systems to handle uncertainty and generalize beyond their training data. AI models consistently fail to make robust enough predictions when facing unfamiliar or adversarial data. Traditional machine learning approaches struggle to address this issue, due to an overemphasis on data fitting, while current uncertainty quantification approaches suffer from serious limitations. This position paper posits a paradigm shift towards epistemic artificial intelligence, emphasizing the need for models to learn from what they know while at the same time acknowledging their ignorance, using the mathematics of second-order uncertainty measures. This approach, which leverages the expressive power of such measures to efficiently manage uncertainty, offers an effective way to improve the resilience and robustness of AI systems, allowing them to better handle unpredictable real-world environments.
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
Decision-Aligned Evaluation of Uncertainty Quantification
Introduces decision-alignment to evaluate uncertainty metrics against downstream decision utilities and proposes prior-weighted proper scoring rules that align better in benchmarks and case studies.
-
What Intermediate Layers Know: Detecting Jailbreaks from Entropy Dynamics
Entropy dynamics across token positions in intermediate layers of LLMs separate jailbreak prompts from benign ones using trend-based features without extra training.
-
Conformalised imprecise inference for robust extrapolation under limited data
A model-agnostic framework combines conformal prediction and imprecise probabilities to yield coverage-valid imprecise predictions that widen under extrapolation and limited data.
-
Statistical inference with belief functions: A survey
The survey reviews the main contributions to learning belief measures from statistical data as an alternative to probability-based inference.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.