Machine Learning and Irresponsible Inference: Morally Assessing the Training Data for Image Recognition Systems
Just as humans can draw conclusions responsibly or irresponsibly, so too can computers. Machine learning systems that have been trained on data sets that include irresponsible judgments are likely to yield irresponsible predictions as outputs. In this paper I focus on a particular kind of inference a computer system might make: identification of the intentions with which a person acted on the basis of photographic evidence. Such inferences are liable to be morally objectionable, because of a way in which they are presumptuous. After elaborating this moral concern, I explore the possibility that carefully procuring the training data for image recognition systems could ensure that the systems avoid the problem. The lesson of this paper extends beyond just the particular case of image recognition systems and the challenge of responsibly identifying a person’s intentions. Reflection on this particular case demonstrates the importance (as well as the difficulty) of evaluating machine learning systems and their training data from the standpoint of moral considerations that are not encompassed by ordinary assessments of predictive accuracy.
KeywordsMachine learning algorithms Image recognition systems Training data Responsible AI judgment Ingrained responsibility Modular responsibility Intention ascription
I am grateful to Andréa Atkins and to attendees of the IACAP 2016 for discussion of these issues. A preliminary exposition of some of the ideas and arguments presented in this chapter appeared in a short essay posted on the website of the Loyola Center for Digital Ethics and Policy (http://www.digitalethics.org/).
- Davidson, Donald. 1984a. Inquiries into truth and interpretation. Oxford: Clarendon Press.Google Scholar
- ———. 1984b. Belief and the basis of meaning. Reprinted in Davidson (1984a): 141–154.Google Scholar
- ———. 2004b. Expressing evaluations. Reprinted in Davidson (2004a): 19–37.Google Scholar
- Dennett, Daniel. 1989a. The intentional stance. Cambridge, MA: MIT Press.Google Scholar
- ———. 1989b. True believers. Reprinted in Dennet (1989a): 13–35.Google Scholar
- Karpathy, Andrej, and Li Fei-Fei. 2014. Deep visual-semantic alignments for generating image descriptions. arXiv preprint arXiv:1412.2306.Google Scholar
- Lippmann, Walter. 1922. Public opinion. New York: Macmillan.Google Scholar
- Marvit, Moshe. 2014. How crowdworkers became the ghosts in the digital machine. The Nation. http://www.thenation.com/article/how-crowdworkers-became-ghosts-digital-machine/. Accessed 11 Jan 2016.
- Park, Eunbyung, Xufeng Han, Tamara Berg, and Alexander Berg. (unpublishedms). Combining multiple sources of knowledge in deep CNNs for action recognition. http://www.cs.unc.edu/~eunbyung/papers/wacv2016_combining.pdf. Accessed 11 Jan 2016.
- Rashtchian, Cyrus, Peter Young, Micah Hodosh, and Julia Hockenmaier. 2010. Collecting image annotations using Amazon’s Mechanical Turk. In Proceedings of the NAACL HLT 2010 Workshop on Creating Speech and Language Data with Amazon’s Mechanical Turk, 139–147. Association for Computational Linguistics.Google Scholar
- Vinyals, Oriol, Alexander Toshev, Samy Bengio, and Dumitru Erhan. 2014. Show and tell: A neural image caption generator. arXiv preprint arXiv:1411.4555.Google Scholar