Establishing Trustworthiness: Rethinking Tasks and Model Evaluation
This work tackles the problem of ensuring reliable and trustworthy NLP systems for real-world deployments, particularly in zero-shot scenarios, but it is incremental as it reviews existing approaches and offers recommendations without introducing new methods.
The paper addresses the challenge of evaluating large language models (LLMs) as traditional task-specific approaches break down, arguing for a holistic view centered on trustworthiness and providing recommendations for multi-faceted evaluation protocols.
Language understanding is a multi-faceted cognitive capability, which the Natural Language Processing (NLP) community has striven to model computationally for decades. Traditionally, facets of linguistic intelligence have been compartmentalized into tasks with specialized model architectures and corresponding evaluation protocols. With the advent of large language models (LLMs) the community has witnessed a dramatic shift towards general purpose, task-agnostic approaches powered by generative models. As a consequence, the traditional compartmentalized notion of language tasks is breaking down, followed by an increasing challenge for evaluation and analysis. At the same time, LLMs are being deployed in more real-world scenarios, including previously unforeseen zero-shot setups, increasing the need for trustworthy and reliable systems. Therefore, we argue that it is time to rethink what constitutes tasks and model evaluation in NLP, and pursue a more holistic view on language, placing trustworthiness at the center. Towards this goal, we review existing compartmentalized approaches for understanding the origins of a model's functional capacity, and provide recommendations for more multi-faceted evaluation protocols.