Abstract
Transparency is often deemed critical to enable effective real-world deployment of intelligent systems. Yet the motivations for and benefits of different types of transparency can vary significantly depending on context, and objective measurement criteria are difficult to identify. We provide a brief survey, suggesting challenges and related concerns, particularly when agents have misaligned interests. We highlight and review settings where transparency may cause harm, discussing connections across privacy, multi-agent game theory, economics, fairness and trust.
Supported by The Alan Turing Institute, Darwin College and the Leverhulme Trust.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Greater faithfulness of an explanation may challenge the ability of its audience to understand it well, perhaps requiring a greater investment of time and effort [54].
- 2.
One view is that if rules are set up correctly, then transparency will not lead to ‘gaming’ since agents optimizing their own objectives subject to the rules will necessarily lead to a good outcome for all. However, it is often very challenging in practice to get the rules exactly right in this way – thus there may be a distinction between the ‘letter’ and the ‘spirit’ of the law. See Sect. 2.4 for a related example.
- 3.
Consider \(c(x)= {\left\{ \begin{array}{ll} 10x &{} x \le 3\\ 30 &{} 3 \le x \le 3+\epsilon \\ 10(x-\epsilon ) &{} 3+\epsilon \le x. \end{array}\right. }\).
References
Adel, T., Ghahramani, Z., Weller, A.: Discovering interpretable representations for both deep generative and discriminative models. In: ICML (2018)
Adelberg, A.: Narrative disclosures contained in financial reports: means of communication or manipulation? Acc. Bus. Res. 9(35), 179–190 (1979)
Amodei, D., Olah, C., Steinhardt, J., Christiano, P., Schulman, J., Mané, D.: Concrete problems in AI safety. arXiv preprint. arXiv:1606.06565 (2016)
Angwin, J., Larson, J., Mattu, S., Kirchner, L.: Machine bias: There’s software used across the country to predict future criminals. and it’s biased against blacks. ProPublica, May 2016
Arrow, K., Debreu, G.: Existence of an equilibrium for a competitive economy. Econometrica J. Econometric Soc. 22, 265–290 (1954)
Axelrod, R.: The Evolution of Cooperation. Basic Books, New York (2006)
Bach, S., Binder, A., Montavon, G., Klauschen, F., Müller, K.R., Samek, W.: On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation. PLoS ONE 10(7), e0130140 (2015)
Barocas, S., Selbst, A.: Big data’s disparate impact. Calif. Law Rev. 104(3), 671–732 (2016)
Bernstein, E.: The transparency trap. Harv. Bus. Rev. 92(10), 58–66 (2014)
Braess, D.: Über ein paradoxon aus der verkehrsplanung. Math. Methods Oper. Res. 12(1), 258–268 (1968)
Chen, X., Duan, Y., Houthooft, R., Schulman, J., Sutskever, I., Abbeel, P.: InfoGAN: interpretable representation learning by information maximizing generative adversarial nets. In: NeurIPS (2016)
Coldewey, D.: Bulgaria now requires (some) government software be open source. TechCrunch, July 2016
Cole, D.: We kill people based on metadata. The New York Review of Books, May 2014
Dabkowski, P., Gal, Y.: Real time image saliency for black box classifiers. In: NeurIPS (2017)
Datta, A., Sen, S., Zick, Y.: Algorithmic transparency via quantitative input influence: theory and experiments with learning systems. In: 2016 IEEE Symposium on Security and Privacy (SP), pp. 598–617. IEEE (2016)
Doshi-Velez, F., Kim, B.: Towards a rigorous science of interpretable machine learning. CoRR abs/1702.08608 (2017)
Dwork, C., Hardt, M., Pitassi, T., Reingold, O.: Fairness through awareness. In: ITCSC (2012)
Einstein, A.: The common language of science. Adv. Sci. 2(5) (1942). A 1941 recording by Einstein to the British Association for the Advancement of Science is available https://www.youtube.com/watch?v=e3B5BC4rhAU
Etzioni, A.: Is transparency the best disinfectant? J. Polit. Philos. 18(4), 389–404 (2010)
Evtimova, K., Drozdov, A., Kiela, D., Cho, K.: Emergent communication in a multi-modal, multi-step referential game. In: ICLR (2018)
Feldman, M., Friedler, S.A., Moeller, J., Scheidegger, C., Venkatasubramanian, S.: Certifying and removing disparate impact. In: KDD (2015)
Ghani, R.: You say you want transparency and interpretability? (2016). http://www.rayidghani.com/you-say-you-want-transparency-and-interpretability
Gilovich, T., Savitsky, K., Medvec, V.: The illusion of transparency: biased assessments of others’ ability to read one’s emotional states. J. Pers. Soc. Psychol. 75(2), 332 (1998)
Grgić-Hlača, N., Zafar, M.B., Gummadi, K.P., Weller, A.: On fairness, diversity and randomness in algorithmic decision making. In: FAT/ML Workshop at KDD (2017)
Griffin, D., Ross, L.: Subjective construal, social inference, and human misunderstanding. Adv. Exp. Soc. Psychol. 24, 319–359 (1991)
Hammond, R., Axelrod, R.: The evolution of ethnocentrism. J. Conflict Resolut. 50(6), 926–936 (2006). http://www-personal.umich.edu/~axe/vtmovie.htm
Hardt, M., Price, E., Srebro, N.: Equality of opportunity in supervised learning. In: NeurIPS (2016)
Havrylov, S., Titov, I.: Emergence of language with multi-agent games: learning to communicate with sequences of symbols. arXiv preprint. arXiv:1705.11192 (2017)
Higgins, I., et al.: beta-VAE: learning basic visual concepts with a constrained variational framework. In: ICLR (2017)
Horling, B., Lesser, V.: A survey of multi-agent organizational paradigms. Knowl. Eng. Rev. 19(4), 281–316 (2004)
Joseph, M., Kearns, M., Morgenstern, J., Roth, A.: Fairness in learning: classic and contextual bandits. In: NeurIPS (2016)
Kelly, F.: Network routing. Philos. Trans. Royal Soc. Lond. A Math. Phys. Eng. Sci. 337(1647), 343–367 (1991)
Kilbertus, N., Gascón, A., Kusner, M.J., Veale, M., Gummadi, K.P., Weller, A.: Blind justice: fairness with encrypted sensitive attributes. In: ICML (2018)
Koh, P.W., Liang, P.: Understanding black-box predictions via influence functions. In: ICML (2017)
Lakkaraju, H., Bach, S., Leskovec, J.: Interpretable decision sets: a joint framework for description and prediction. In: KDD, pp. 1675–1684. ACM (2016)
Lakonishok, J., Shleifer, A., Vishny, R.W., Hart, O., Perry, G.L.: The structure and performance of the money management industry. Brook. Pap. Econ. Act. Microecon. 1992, 339–391 (1992)
Langer, E., Blank, A., Chanowitz, B.: The mindlessness of ostensibly thoughtful action: the role of “placebic” information in interpersonal interaction. J. Pers. Soc. Psychol. 36(6), 635–642 (1978)
Lawrence, N.: Living together: mind and machine intelligence. CoRR abs/1705. 07996 (2017)
Leibo, J., Zambaldi, V., Lanctot, M., Marecki, J., Graepel, T.: Multi-agent reinforcement learning in sequential social dilemmas. In: AAMAS (2017). https://deepmind.com/blog/understanding-agent-cooperation/
Lerner, J., Tetlock, P.: Accounting for the effects of accountability. Psychol. Bull. 125(2), 255 (1999)
Levine, E., Schweitzer, M.: Prosocial lies: when deception breeds trust. Organ. Behav. Hum. Decis. Process. 126, 88–106 (2015)
Lipton, Z.: The mythos of model interpretability. In: ICML Workshop on Human Interpretability in Machine Learning, New York, NY, pp. 96–100 (2016)
Lowe, R., Wu, Y., Tamar, A., Harb, J., Abbeel, P., Mordatch, I.: Multi-agent actor-critic for mixed cooperative-competitive environments. In: NeurIPS (2017). https://blog.openai.com/learning-to-cooperate-compete-and-communicate/
Lowenstein, L.: Financial transparency and corporate governance: you manage what you measure. Columbia Law Rev. 96(5), 1335–1362 (1996)
Lundberg, S.M., Lee, S.I.: A unified approach to interpreting model predictions. In: NeurIPS (2017)
McAllister, R., et al.: Concrete problems for autonomous vehicle safety: advantages of Bayesian deep learning. In: IJCAI (2017)
Mordatch, I., Abbeel, P.: Emergence of grounded compositional language in multi-agent populations. In: AAAI (2018)
Mortier, R., Haddadi, H., Henderson, T., McAuley, D., Crowcroft, J.: Human-data interaction: the human face of the data-driven society. Social Science Research Network (2014)
O’Neill, O.: What we don’t understand about trust. TED talk, June 2013
Pandey, G.: Indian Supreme Court in landmark ruling on privacy. BBC News, 24 August 2017. http://www.bbc.co.uk/news/world-asia-india-41033954
Peled, A.: When transparency and collaboration collide: the USA open data program. J. Am. Soc. Inf. Sci. Technol. 62(11), 2085–2094 (2011)
Peters, J., Janzing, D., Schölkopf, B.: Elements of Causal Inference - Foundations and Learning Algorithms. Adaptive Computation and Machine Learning Series. MIT Press, Cambridge (2017)
Prat, A.: The wrong kind of transparency. Am. Econ. Rev. 95(3), 862–877 (2005)
Ribeiro, M., Singh, S., Guestrin, C.: Why should I trust you?: Explaining the predictions of any classifier. In: KDD (2016)
Ross, S.A.: The economic theory of agency: the principal’s problem. Am. Econ. Rev. 63(2), 134–139 (1973)
Roughgarden, T.: Selfish routing and the price of anarchy. Technical report, Stanford University Dept of Computer Science (2006). http://www.dtic.mil/get-tr-doc/pdf?AD=ADA637949
Rudin, C.: Please stop explaining black box models for high stakes decisions. In: NeurIPS Workshop on Critiquing and Correcting Trends in Machine Learning (2018)
Rudin, C., Ustun, B.: Optimized scoring systems: toward trust in machine learning for healthcare and criminal justice. Interfaces 48(5), 449–466 (2018)
Samek, W., Binder, A., Montavon, G., Lapuschkin, S., Müller, K.R.: Evaluating the visualization of what a deep neural network has learned. IEEE Trans. Neural Networks Learn. Syst. 28(11), 2660–2673 (2017)
Shapley, L.S.: A value for n-person games. Contrib. Theory Games 2(28), 307–317 (1953)
Solove, D.: Privacy and power: computer databases and metaphors for information privacy. Stanford Law Rev. 53(6), 1393–1462 (2001)
Solove, D.: Why privacy matters even if you have ‘nothing to hide’. Chronicle of Higher Education, 15 May 2011
Sridhar, D., Batniji, R.: Misfinancing global health: a case for transparency in disbursements and decision making. Lancet 372(9644), 1185–1191 (2008)
Štrumbelj, E., Kononenko, I.: Explaining prediction models and individual predictions with feature contributions. Knowl. Inf. Syst. 41(3), 647–665 (2014)
Sundararajan, M., Taly, A., Yan, Q.: Axiomatic attribution for deep networks. In: ICML (2017)
Tishby, N., Pereira, F., Bialek, W.: The information bottleneck method. In: Allerton Conference on Communication, Control, and Computing (1999)
Varshney, K.R., Alemzadeh, H.: On the safety of machine learning: cyber-physical systems, decision sciences, and data products. Big Data 5(3), 246–255 (2017)
Varshney, L., Varshney, K.: Decision making with quantized priors leads to discrimination. Proc. IEEE 105(2), 241–255 (2017)
Vishwanath, T., Kaufmann, D.: Toward transparency: new approaches and their application to financial markets. World Bank Res. Obs. 16(1), 41–57 (2001)
Wachter, S., Mittelstadt, B., Russell, C.: Counterfactual explanations withoutopening the black box: automated decisions and the GDPR. Harv. J. Law Technol. 31(2), 841 (2018)
Weller, A.: Challenges for transparency. In: ICML Workshop on Human Interpretability (2017)
Wiggers, K.: New York City announces task force to find biases in algorithms. VentureBeat, May 2018
Wiley, R.: The evolution of communication: information and manipulation. Anim. Behav. 2, 156–189 (1983)
Zafar, M.B., Valera, I., Rodriguez, M.G., Gummadi, K.P., Weller, A.: From parity to preference-based notions of fairness in classification. In: NeurIPS (2017)
Zintgraf, L., Cohen, T., Adel, T., Welling, M.: Visualizing deep neural network decisions: prediction difference analysis. In: ICLR (2017)
Acknowledgements
This article is an extended version of [71]. The author thanks Frank Kelly for pointing out the Braess’ paradox example and related intuition; and thanks Vasco Carvalho, Stephen Cave, Jon Crowcroft, David Fohrman, Yarin Gal, Adria Gascon, Zoubin Ghahramani, Sanjeev Goyal, Krishna P. Gummadi, Dylan Hadfield-Menell, Bill Janeway, Frank Kelly, Aryeh Kontorovich, Neil Lawrence, Barney Pell and Mark Rowland for helpful discussions; and thanks the anonymous reviewers for helpful comments. The author acknowledges support from the David MacKay Newton research fellowship at Darwin College, The Alan Turing Institute under EPSRC grant EP/N510129/1 & TU/B/000074, and the Leverhulme Trust via the CFI.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Weller, A. (2019). Transparency: Motivations and Challenges. In: Samek, W., Montavon, G., Vedaldi, A., Hansen, L., Müller, KR. (eds) Explainable AI: Interpreting, Explaining and Visualizing Deep Learning. Lecture Notes in Computer Science(), vol 11700. Springer, Cham. https://doi.org/10.1007/978-3-030-28954-6_2
Download citation
DOI: https://doi.org/10.1007/978-3-030-28954-6_2
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-28953-9
Online ISBN: 978-3-030-28954-6
eBook Packages: Computer ScienceComputer Science (R0)