While demands for change and accountability for harmful AI consequences
...
Even when aggregate accuracy is high, state-of-the-art NLP models often ...
Despite substantial advancements, Natural Language Processing (NLP) mode...
Large language models are becoming increasingly pervasive and ubiquitous...
Artificial intelligence (AI) researchers have been developing and refini...
Large language models (LLMs) can perform complex reasoning in few- and
z...
The in-context learning capabilities of LLMs like GPT-3 allow annotators...
Changing how pre-trained models behave – e.g., improving their performan...
Vision models often fail systematically on groups of data that share com...
Current approaches for fixing systematic problems in NLP models (e.g. re...
We introduce SpotCheck, a framework for generating synthetic datasets to...
Interpretability methods are developed to understand the working mechani...
Machine learning models often use spurious patterns such as "relying on ...
Counterfactual examples have been shown to be useful for many applicatio...
Increasingly, organizations are pairing humans with AI systems to improv...
Although measuring held-out accuracy has been the primary approach to
ev...
Recent work in model-agnostic explanations of black-box machine learning...
At the core of interpretable machine learning is the question of whether...
Understanding why machine learning models behave the way they do empower...
Despite widespread adoption, machine learning models remain mostly black...