Value-laden disciplinary shifts in machine learning

By Ravit Dotan, Smitha Milli. 2019

55 citations recorded by the source corpus.

As machine learning models are increasingly used for high-stakes decision making, scholars have sought to intervene to ensure that such models do not encode undesirable social and political values. However, little attention thus far has been given to how values influence the machine learning discipline as a whole. How do values influence what the discipline focuses on and the way it develops? If undesirable values are at play at the level of the discipline, then intervening on particular models will not suffice to address the problem. Instead, interventions at the disciplinary-level are required. This paper analyzes the discipline of machine learning through the lens of philosophy of science. We develop a conceptual framework to evaluate the process through which types of machine learning models (e.g. neural networks, support vector machines, graphical models) become predominant. The rise and fall of model-types is often framed as objective progress. However, such disciplinary shifts are more nuanced. First, we argue that the rise of a model-type is self-reinforcing-it influences the way model-types are evaluated. For example, the rise of deep learning was entangled with a greater focus on evaluations in compute-rich and data-rich environments. Second, the way model-types are evaluated encodes loaded social and political values. For example, a greater focus on evaluations in compute-rich and data-rich environments encodes values about centralization of power, privacy, and environmental concerns.

Subfields: Safety, Robustness & Interpretability

How researchers see this paper

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

In Safety, Robustness & Interpretability

Citations in the corpus

AI and the Everything in the Whole Wide World Benchmark (2021, background): “Dotan and Milli discuss these benchmarks as value laden and definitive of the algorithmic approaches that end up defining the field [Dotan and Milli, 2020], with several scholars agreeing to the important role such benchmarks play in determining the nature of algorithmic development and in many case”
Data and its (dis)contents: A survey of dataset development and use in machine learning research (2020, background): “Benchmark datasets have also played a critical role in orienting the goals, values, and research agendas of the machine learning community.(5) In recent years, machine learning systems have been reported to achieve ‘‘super-human’’ performance when evaluated on such benchmark datasets.”
Bringing the People Back In: Contesting Benchmark Machine Learning Datasets (2020, background): “Datasets that have achieved such authoritative status also play a unique and powerful role in structuring research agendas and values within machine learning subﬁelds (Dotan & Milli, 2020).”
The Values Encoded in Machine Learning Research (2021, background): “Dotan and Milli [20] argue that the rise of deep learning is value-laden, promoting the centralization of power among other political values.”
On the genealogy of machine learning datasets: A critical history of ImageNet (2021, background): “This marked the ﬁ rst use of neutral networks in the challenge and is often credited with sparking the resurgence of neural networks (under a new moniker — deep learning) as a dominant ML paradigm (Dotan and Milli, 2020).”
Image Cropping on Twitter: Fairness Metrics, their Limitations, and the Importance of Representation, Design, and Agency (2021, background): “Additionally, there have been many calls for a more critical evaluation of the underlying values and assumptions embedded in machine learning and human computer interaction (HCI) systems [4, 12, 28, 36, 37, 77].”
Dynaboard: An Evaluation-As-A-Service Platform for Holistic Next-Generation Benchmarking (2021, background): “This gives rise to a “leaderboard culture” in which accuracy is the only thing that matters, even in contexts in which other pressures – e.g., compactness, fairness, efﬁciency – are clearly important [60, 6, 16, 63, 68, 38, 2, 51, 17].”
Evaluation Examples are not Equally Informative: How should that change NLP Leaderboards? (2021, background): “, 2020), and thoughtful consideration of their influence on the direction of research (Sculley et al., 2018; Dotan and Milli, 2020).”
Reduced, Reused and Recycled: The Life of a Dataset in Machine Learning Research (2021, background): “Today, benchmarking is the dominant paradigm for scientiﬁc evaluation in MLR, and the ﬁeld collectively views upward trends on benchmarks as noisy but meaningful indicators of scientiﬁc progress [1, 2, 17].”