On Ensuring that Intelligent Machines Are Well-Behaved
By Philip S. Thomas, Bruno Castro da Silva, Andrew Barto, Emma Brunskill. 2017
16 citations recorded by the source corpus.
Machine learning algorithms are everywhere, ranging from simple data analysis and pattern recognition tools used across the sciences to complex systems that achieve super-human performance on various tasks. Ensuring that they are well-behaved---that they do not, for example, cause harm to humans or act in a racist or sexist way---is therefore not a hypothetical problem to be dealt with in the future, but a pressing one that we address here. We propose a new framework for designing machine learning algorithms that simplifies the problem of specifying and regulating undesirable behaviors. To show the viability of this new framework, we use it to create new machine learning algorithms that preclude the sexist and harmful behaviors exhibited by standard machine learning algorithms in our experiments. Our framework for designing machine learning algorithms simplifies the safe and responsible application of machine learning.
Subfields: Artificial Intelligence
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
- Gabriel Dulac-Arnold
- Cosmin Paduraru
- Jerry Zheng Li
- Sven Gowal
- Nir Levine
- Todd Hester
- Daniel Jaymin Mankowitz
- Michael O'Neill
Citations in the corpus
- Safe Policy Improvement with Baseline Bootstrapping (2017, background): “But in return, it is inherently safe as it only allows to search in a set of policies for which the improvement over the baseline can be safely evaluated (Thomas et al., 2017).”
- Themis: automatically testing software for discrimination (2018, background): “Reducing discrimination in machine learning classifiers [2, 17, 26, 30, 31, 51, 56, 58] and selection algorithms [47] is important work that is complementary to ours.”
- Automatic programming: The open issue? (2019, background): “One approach is to shift responsibility towards the designer of the intelligent algorithm to ensure that it is well behaved [63].”
- Challenges of real-world reinforcement learning: definitions, benchmarks and analysis (2021, background): “This D4PG variant learns the learning rate of the lagrange multiplier in a soft-constrained optimization procedure.”