Evaluation Gaps in Machine Learning Practice

By Ben Hutchinson, Negar Rostamzadeh, Christina Greer, Katherine A. Heller, Vinodkumar Prabhakaran. 2022

101 citations recorded by the source corpus.

Forming a reliable judgement of a machine learning (ML) model’s appropriateness for an application ecosystem is critical for its responsible use, and requires considering a broad range of factors including harms, benefits, and responsibilities. In practice, however, evaluations of ML models frequently focus on only a narrow range of decontextualized predictive behaviours. We examine the evaluation gaps between the idealized breadth of evaluation concerns and the observed narrow focus of actual evaluations. Through an empirical study of papers from recent high-profile conferences in the Computer Vision and Natural Language Processing communities, we demonstrate a general focus on a handful of evaluation methods. By considering the metrics and test data distributions used in these methods, we draw attention to which properties of models are centered in the field, revealing the properties that are frequently neglected or sidelined during evaluation. By studying these properties, we demonstrate the machine learning discipline’s implicit assumption of a range of commitments which have normative impacts; these include commitments to consequentialism, abstractability from context, the quantifiability of impacts, the limited role of model inputs in evaluation, and the equivalence of different failure modes. Shedding light on these assumptions enables us to question their appropriateness for ML system contexts, pointing the way towards more contextualized evaluation methodologies for robustly examining the trustworthiness of ML models.

Subfields: Artificial Intelligence, Safety Research, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

  1. Vinodkumar Prabhakaran
  2. Ben Hutchinson
  3. Su Lin Blodgett
  4. Irene Solaiman
  5. Dylan Baker
  6. Alexandra Sasha Luccioni
  7. Zeerak Talat
  8. Margaret Mitchell

In Safety Research

  1. Irene Solaiman
  2. Ben Hutchinson
  3. Vinodkumar Prabhakaran
  4. Marie-Therese Png
  5. Zeerak Talat
  6. Yacine Jernite
  7. Alexandra Sasha Luccioni
  8. Su Lin Blodgett

In Evaluation & Benchmarks

  1. Ben Hutchinson
  2. Irene Solaiman
  3. Zeerak Talat
  4. Yacine Jernite
  5. Alexandra Sasha Luccioni
  6. Vinodkumar Prabhakaran
  7. Jesse Dodge
  8. Su Lin Blodgett

Citations in the corpus