本文的重点是先进的计算系统和人工智能如何具体促进新的科学理解。有许多相关的、有趣的话题,我们在这里无法涉及。例如,我们将不讨论科学理解与认知科学之间的关系,而是向读者推荐一个很好的概述[14]。此外,我们将只讨论自然科学背景下的“理解”,在自然科学中我们可以使用科学哲学中的具体标准,因此,我们不会触及更广泛背景下的“理解”(比如婴儿和动物的理解,人工智能中的语义理解以及相关话题)。许多其他作品对相关问题有所贡献,在此应该提及。人工智能的一个重要研究领域是可解释人工智能,其目的是解释和说明先进的人工智能算法如何得出它们的解决方案;例如,见参考文献[15-18]。虽然没有必要,而且我们认为也不足以解释人工智能的内部运作来获得新的科学理解,但许多这些工具和技术可能非常有用。我们将在下面用自然科学中的具体例子来简要解释它们。 人工智能先驱 Donald Michie 将机器学习(ML)分为三类:弱人工智能、强人工智能和超人工智能,其中超人工智能需要机器来教人[19]。超人工智能的机器学习与理解主体的概念有关,我们将在下文中进行定义和详细介绍。一个非常有用的、可理解的科学计算和人工智能方法的集合可以在参考文献[20]中找到。参考文献[21]中描述了分子设计的不同自动化水平,最后一步是由计算机来选择最初的想法。其他作品研究了基于特定科学方法的完全自动化可能是什么样子,从而产生了“诺贝尔图灵挑战”[22]的想法,即开发一个能够做出诺贝尔奖级别科学发现的人工智能系统。我们注意到,我们的观点有目的地不依赖于任何特定的科学方法(以避免基础层面的问题[23])。相反,我们专注于科学家如何获得“科学理解”,以及先进的人工智能如何帮助人类获得新的科学理解。
2. 科学理解
想象一个先知可以提供永远正确的重要预测。尽管这样一个假设的存在会产生重大的科学影响,但科学家们并不满足。他们希望“能够掌握预测是如何产生的,并对具体情形下的后果产生感觉”[13]。通俗地说,我们把这个目标称为“理解”,但这到底是什么意思呢?为了找到科学理解的标准,我们从科学哲学中寻求指导。尽管几乎没有科学家会反对将“理解”作为科学的基本目标(与解释、描述和预测并列[24]),但这种观点并不总是被哲学家所接受。卡尔·亨普尔(Carl Hempel)对澄清“科学解释”的含义做出了基础性的贡献,他认为“理解”是主观的,只是科学活动的心理副产品,因此与科学哲学无关[25]。众多哲学家批评了这一结论,试图正式确定“科学理解”的实际含义。这些提议表明,“理解“与建立因果模型的能力有关(例如,开尔文勋爵说:“在我看来,‘我们是否理解物理学中的某一主题’的检验标准是,‘我们能否为它建立一个机械模型?’”[13]),与提供可视化(或明确表达,正如其坚定的支持者埃尔温·薛定谔所称[26,27])有关,或者理解对应于提供思想的统一性[28,29]。最近,Henk de Regt 和 Dennis Dieks 发展了一种新的科学理解理论,它既是背景性的,也是实用性的[12,13,24]。他们发现,诸如可视化或统一化的技术是“理解的工具”,从而将以前的想法连接到一个总体框架中。他们的理论对正在使用的具体“工具”是不可知的,这使得它在各种科学学科中的应用特别有用。de Regt 和 Dieks 扩展了海森堡(Werner Heisenberg)的见解[30],他们的理论背后的主要动机是“令人满意的科学理解概念应该反映科学的实际(当代和历史)实践”,而不是仅仅引入理论或假设的想法。简单地说,他们认为“如果存在一个关于 P 的可理解的理论 T,使科学家能够在不进行精确计算的情况下认识到T的定性特征后果,那么一个现象 P 就可以被理解”[12,13]。de Regt 和 Dieks 定义了两个相互关联的标准。
理解现象的标准:如果存在一个关于 P 的可理解的理论 T,那么现象 P 就可以被理解。
理论的可理解性标准:一个科学理论 T 对科学家来说是可理解的(在背景 C 中),如果他们能够认识到T的定性特征后果而不进行精确计算。
在计算机辅助理解的第一个维度中,人类科学家应该对来自计算显微镜的新数据进行概括。因此,我们认为数据表示的进步可以极大地帮助人类掌握基础结构,促进新的科学理解。科学家们目前主要是在(可能是动画的)二维图形表示中分析数据。我们相信,真正的 3D 环境(通过虚拟或增强现实眼镜或全息技术实现)将大大有助于对复杂系统或复杂数据的理解。这方面的初步进展已经在化学[49-51]和天体物理学[52]中得到证明,我们期望这将成为科学家的标准工具。此外,时间维度可以用来表示更多的结构化数据;例如,通过视频( 3D 视频)。另外,声音也可以作为一个额外的数据维度,因为人类的听觉在检测(周期性)时间相关数据的结构或对称性方面非常出色。这个机会已经在高能物理学[53]和天文学[54]的几十个项目中得到了探索。一个强大的算法也许能够识别基础数据中的对称性,并将其投射到带有声音的三维视频中,这可能有助于人类识别并随后理解计算显微镜所产生的数据中的新属性。
1. Zdeborová, L. New tool in the box. Nat. Phys.13, 420–421 (2017).2. Fösel, T., Tighineanu, P., Weiss, T. & Marquardt, F. Reinforcement learning with neural networks for quantum feedback. Phys. Rev. X8, 031084 (2018).3. Melnikov, A. A. et al. Active learning machine learns to create new quantum experiments. Proc. Natl Acad. Sci. USA115, 1221–1226 (2018).4. Aspuru-Guzik, A., Lindh, R. & Reiher, M. The matter simulation (r)evolution. ACS Cent. Sci.4, 144–152 (2018).5. Hoffmann, R. & Malrieu, J.-P. Simulation vs. understanding: a tension, in quantum chemistry and beyond. Part A. Stage setting. Angew. Chem.132, 12690–12710 (2020).6. Hoffmann, R. & Malrieu, J.-P. Simulation vs. understanding: a tension, in quantum chemistry and beyond. Part B. The march of simulation, for better or worse. Angew. Chem. Int. Ed.59, 13156–13178 (2020).7. Hoffmann, R. & Malrieu, J.-P. Simulation vs. understanding: a tension, in quantum chemistry and beyond. Part C. Toward consilience. Angew. Chem. Int. Ed.59, 13694–13710 (2020).8. Marcus, G. The next decade in AI: four steps towards robust artificial intelligence. Preprint at arXiv 2002.06177 (2020).9. Thaler, J. Designing an AI physicist. CERN Courier, https://cerncourier.com/a/designing-an-ai-physicist/ (2021).10. Potochnik, A. The diverse aims of science. Stud. Hist. Philos. Sci. A53, 71–80 (2015).11. Potochnik, A. Idealization and the Aims of Science (Univ. Chicago Press, 2017).12. de Regt, H. W. Understanding Scientific Understanding (Oxford Univ. Press, 2017).13. De Regt, H. W. & Dieks, D. A contextual approach to scientific understanding. Synthese144, 137–170 (2005).14. Boden, M. A. Mind as Machine: A History of Cognitive Science (Oxford Univ. Press, 2008).15. Doran, D., Schulz, S. & Besold, T. R. What does explainable AI really mean? A new conceptualization of perspectives. Preprint at arXiv 1710.00794 (2017).16. Tjoa, E. & Guan, C. A survey on explainable artificial intelligence (XAI): toward medical XAI. IEEE Trans. Neural Netw. Learn. Syst.32, 4793–4813 (2020).17. Burkart, N. & Huber, M. F. A survey on the explainability of supervised machine learning. J. Artif. Intell. Res.70, 245–317 (2021).18. Samek, W., Montavon, G., Vedaldi, A., Hansen, L. K. & Müller, K.-R. Explainable AI: Interpreting, Explaining and Visualizing Deep Learning (Springer, 2019).19. Michie, D. in Proc. 3rd European Conference on European Working Session on Learning, 107–122 (ACM, 1988).20. Lavin, A. et al. Simulation intelligence: Towards a new generation of scientific methods. Preprint at arXiv 2112.03235 (2021).21. Goldman, B., Kearnes, S., Kramer, T., Riley, P. & Walters, W. P. Defining levels of automated chemical design. J. Med. Chem.65, 7073–7087 (2022).22. Kitano, H. Nobel Turing Challenge: creating the engine for scientific discovery. NPJ Syst. Biol. Appl.7, 29 (2021).23. Feyerabend, P. Against Method (Verso, 1993).24. De Regt, H. W. Understanding, values, and the aims of science. Philos. Sci.87, 921–932 (2020).25. Hempel, C. G. Aspects of Scientific Explanation (Free Press, 1965).26. Schrödinger, E. ‘Nature and the Greeks’ and ‘Science and Humanism’ (Cambridge Univ. Press, 1996).27. De Regt, H. W. Visualization as a tool for understanding. Perspect. Sci.22, 377–396 (2014).28. Friedman, M. Explanation and scientific understanding. J. Philos.71, 5–19 (1974).29. Kitcher, P. Explanatory unification. Philos. Sci.48, 507–531 (1981).30. Heisenberg, W. Über den anschaulichen Inhalt der quantentheoretischen Kinematik und Mechanik. Z. Phys.43, 172–198 (1927).31. Gómez-Bombarelli, R. et al. Design of efficient molecular organic light-emitting diodes by a high-throughput virtual screening and experimental approach. Nat. Mater.15, 1120–1127 (2016).32. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature596, 583–589 (2021).33. Tunyasuvunakool, K. et al. Highly accurate protein structure prediction for the human proteome. Nature596, 590–596 (2021).34. Iten, R., Metger, T., Wilming, H., Del Rio, L. & Renner, R. Discovering physical concepts with neural networks. Phys. Rev. Lett.124, 010508 (2020).35. Seif, A., Hafezi, M. & Jarzynski, C. Machine learning the thermodynamic arrow of time. Nat. Phys.17, 105–113 (2021).36. Udrescu, S.-M. & Tegmark, M. AI Feynman: A physics-inspired method for symbolic regression. Sci. Adv.6, eaay2631 (2020).37. Lemos, P., Jeffrey, N., Cranmer, M., Ho, S. & Battaglia, P. Rediscovering orbital mechanics with machine learning. Preprint at arXiv 2202.02306 (2022).38. Casalino, L. et al. Beyond shielding: the roles of glycans in the SARS-CoV-2 spike protein. ACS Cent. Sci.6, 1722–1734 (2020).39. Fogarty, C. A., Harbison, A. M., Dugdale, A. R. & Fadda, E. How and why plants and human n-glycans are different: Insight from molecular dynamics into the “glycoblocks” architecture of complex carbohydrates. Beilstein J. Org. Chem.16, 2046–2056 (2020).40. Friederich, P., Häse, F., Proppe, J. & Aspuru-Guzik, A. Machine-learned potentials for next-generation matter simulations. Nat. Mater.20, 750–761 (2021).41. Gigan, S., Krzakala, F., Daudet, L. & Carron, I. Artificial intelligence: from electronics to optics. Photoniques104, 49–52 (2020).42. Xu, X. et al. 11 TOPS photonic convolutional accelerator for optical neural networks. Nature589, 44–51 (2021).43. Quantum, G. A. et al. Hartree-Fock on a superconducting qubit quantum computer. Science369, 1084–1089 (2020).44. Zhang, J. et al. Observation of a discrete time crystal. Nature543, 217–220 (2017).45. Schweizer, C. et al. Floquet approach to Z2 lattice gauge theories with ultracold atoms in optical lattices. Nat. Phys.15, 1168–1173 (2019).46. Martinez, E. A. et al. Real-time dynamics of lattice gauge theories with a few-qubit quantum computer. Nature534, 516–519 (2016).47. Cao, Y. et al. Quantum chemistry in the age of quantum computing. Chem. Rev.119, 10856–10915 (2019).48. Gross, C. & Bloch, I. Quantum simulations with ultracold atoms in optical lattices. Science357, 995–1001 (2017).49. O’Connor, M. et al. Sampling molecular conformations and dynamics in a multiuser virtual reality framework. Sci. Adv.4, eaat2731 (2018).50. Probst, D. & Reymond, J.-L. Exploring DrugBank in virtual reality chemical space. J. Chem. Inf. Model.58, 1731–1735 (2018).51. Schmid, J. R., Ernst, M. J. & Thiele, G. Structural chemistry 2.0: combining augmented reality and 3D online models. J. Chem. Educ.97, 4515–4519 (2020).52. Foley, M. et al. A 3D view of Orion: I. Barnard’s Loop. Authorea. https://doi.org/10.22541/au.165540488.82174026/v1 (2022).53. Hill, E., Cherston, J., Goldfarb, S. & Paradiso, J. A. in Proc. 38th International Conference on High Energy Physics, 1042 (2016).54. Zanella, A. et al. Sonification and sound design for astronomy research, education and public engagement. Nat. Astron.https://doi.org/10.1038/s41550-022-01721-z (2022).55. Turing, A. M. Computing machinery and intelligence. Mind50, 433–460 (1950).56. Lehman, J. et al. The surprising creativity of digital evolution: a collection of anecdotes from the evolutionary computation and artificial life research communities. Artif. Life26, 274–306 (2020).57. Pickard, C. J. & Needs, R. Ab initio random structure searching. J. Phys. Condens. Matter23, 053201 (2011).58. Krenn, M., Malik, M., Fickler, R., Lapkiewicz, R. & Zeilinger, A. Automated search for new quantum experiments. Phys. Rev. Lett.116, 090405 (2016).59. Krenn, M., Erhard, M. & Zeilinger, A. Computer-inspired quantum experiments. Nat. Rev. Phys.2, 649–661 (2020).60. Pickard, C. J. & Needs, R. Highly compressed ammonia forms an ionic crystal. Nat. Mater.7, 775–779 (2008).61. Krenn, M., Hochrainer, A., Lahiri, M. & Zeilinger, A. Entanglement by path identity. Phys. Rev. Lett.118, 080401 (2017).62. Krenn, M., Gu, X. & Zeilinger, A. Quantum experiments and graphs: multiparty states as coherent superpositions of perfect matchings. Phys. Rev. Lett.119, 240403 (2017).63. Krenn, M., Kottmann, J., Tischler, N. & Aspuru-Guzik, A. Conceptual understanding through efficient automated design of quantum optical experiments. Phys. Rev. X11, 031044 (2021).64. Malhotra, P., Vig, L., Shroff, G. & Agarwal, P. in Proc. European Symposium on Artificial Neural Networks (ESANN), 89–94 (2015).65. Collaboration, A. Dijet resonance search with weak supervision using s√=13TeV pp collisions in the ATLAS detector. Phys. Rev. Lett.125, 131801 (2020).66. Collaboration, C. Probing effective field theory operators in the associated production of top quarks with a Z boson in multilepton final states at s√=13TeV. J. High Energy Phys.2021, 83 (2021).67. Park, S. E., Rankin, D., Udrescu, S.-M., Yunus, M. & Harris, P. Quasi anomalous knowledge: searching for new physics with embedded knowledge. J. High Energy Phys.2021, 30 (2021).68. Karagiorgi, G., Kasieczka, G., Kravitz, S., Nachman, B. & Shih, D. Machine learning in the search for new fundamental physics. Nat. Rev. Phys.4, 399–412 (2022).69. Schwartz, M. D. Modern machine learning and particle physics. Preprint at arXiv 2103.12226 (2021).70. Kasieczka, G. et al. The LHC Olympics 2020: a community challenge for anomaly detection in high energy physics. Rep. Prog. Phys.84, 124201 (2021).71. Dawid, A., Huembeli, P., Tomza, M., Lewenstein, M. & Dauphin, A. Hessian-based toolbox for reliable and interpretable machine learning in physics. Mach. Learn. Sci. Technol.3, 015002 (2021).72. Koh, P. W. & Liang, P. in Proc. 34th International Conference on Machine Learning, 1885–1894 (PMLR, 2017).73. Yu, H., Mineyev, I. & Varshney, L. R. A group-theoretic approach to computational abstraction: Symmetry-driven hierarchical clustering. Preprint at arXiv 1807.11167 (2018).74. Dehmamy, N., Walters, R., Liu, Y., Wang, D. & Yu, R. Automatic symmetry discovery with lie algebra convolutional network. Adv. Neural Inf. Process. Syst.34, 2503–2515 (2021).75. Nigam, A. et al. Assigning confidence to molecular property prediction. Expert Opin. Drug Discov.16, 1009–1023 (2021).76. Davies, A. et al. Advancing mathematics by guiding human intuition with AI. Nature 600, 70–74 (2021).77. Douglas, M. R. Machine learning as a tool in theoretical science. Nat. Rev. Phys.4, 145–146 (2022).78. King, R. D. et al. The automation of science. Science324, 85–89 (2009).79. Bédard, A.-C. et al. Reconfigurable system for automated optimization of diverse chemical reactions. Science361, 1220–1225 (2018).80. Steiner, S. et al. Organic synthesis in a modular robotic system driven by a chemical programming language. Science363, eaav2211 (2019).81. Coley, C. W. A robotic platform for flow synthesis of organic compounds informed by AI planning. Science365, eaax1566 (2019).82. Burger, B. et al. A mobile robotic chemist. Nature583, 237–241 (2020).83. Chatterjee, S., Guidi, M., Seeberger, P. H. & Gilmore, K. Automated radial synthesis of organic molecules. Nature579, 379–384 (2020).84. Grizou, J., Points, L. J., Sharma, A. & Cronin, L. A curious formulation robot enables the discovery of a novel protocell behavior. Sci. Adv.6, eaay4237 (2020).85. Moon, H. et al. Machine learning enables completely automatic tuning of a quantum device faster than human experts. Nat. Commun.11, 4161 (2020).86. Dalgaard, M., Motzoi, F., Sørensen, J. J. & Sherson, J. Global optimization of quantum dynamics with AlphaZero deep exploration. NPJ Quantum Inf.6, 6 (2020).87. Larsen, P. & Von Ins, M. The rate of growth in scientific publication and the decline in coverage provided by Science Citation Index. Scientometrics84, 575–603 (2010).88. Reisz, N. et al. Loss of sustainability in scientific work. New J. Phys.24, 053041 (2022).89. Evans, J. A. & Foster, J. G. Metaknowledge. Science331, 721–725 (2011).90. Clauset, A., Larremore, D. B. & Sinatra, R. Data-driven predictions in the science of science. Science355, 477–480 (2017).91. Fortunato, S. et al. Science of science. Science359, eaao0185 (2018).92. Wang, D. & Barabási, A.-L. The Science of Science (Cambridge Univ. Press, 2021).93. Tshitoyan, V. et al. Unsupervised word embeddings capture latent knowledge from materials science literature. Nature571, 95–98 (2019).94. Olivetti, E. A. et al. Data-driven materials research enabled by natural language processing and information extraction. Appl. Phys. Rev.7, 041317 (2020).95. Schwalbe-Koda, D., Jensen, Z., Olivetti, E. & Gómez-Bombarelli, R. Graph similarity drives zeolite diffusionless transformations and intergrowth. Nat. Mater.18, 1177–1181 (2019).96. Rzhetsky, A., Foster, J. G., Foster, I. T. & Evans, J. A. Choosing experiments to accelerate collective discovery. Proc. Natl Acad. Sci. USA112, 14569–14574 (2015).97. Krenn, M. & Zeilinger, A. Predicting research trends with semantic and neural networks with an application in quantum physics. Proc. Natl Acad. Sci. USA117, 1910–1916 (2020).98. Devlin, J., Chang, M.-W., Lee, K. & Toutanova, K. Bert: Pre-training of deep bidirectional transformers for language understanding. Preprint at arXiv 1810.04805 (2018).99. Brown, T. B. et al. Language models are few-shot learners. Preprint at arXiv 2005.14165 (2020).100. Hamilton, W. L., Ying, R. & Leskovec, J. in Proc. 31st International Conference on Neural Information Processing Systems (NIPS 2017), 1025–1035 (2017).101. Montavon, G., Samek, W. & Müller, K.-R. Methods for interpreting and understanding deep neural networks. Digit. Signal Process.73, 1–15 (2018).102. Roscher, R., Bohn, B., Duarte, M. F. & Garcke, J. Explainable machine learning for scientific insights and discoveries. IEEE Access8, 42200–42216 (2020).103. Lundberg, S. M. et al. From local explanations to global understanding with explainable AI for trees. Nat. Mach. Intell.2, 56–67 (2020).104. Mahendran, A. & Vedaldi, A. in Proc. IEEE Conference on Computer Vision and Pattern Recognition, 5188–5196 (2015).105. Mordvintsev, A., Olah, C. & Tyka, M. Inceptionism: going deeper into neural networks. https://ai.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html (2015).106. Shen, C., Krenn, M., Eppel, S. & Aspuru-Guzik, A. Deep molecular dreaming: Inverse machine learning for de-novo molecular design and interpretability with surjective representations. Mach. Learn. Sci. Technol.2, 03LT02 (2021).107. Burgess, C. P. et al. Understanding disentangling in β-VAE. Preprint at arXiv 1804.03599 (2018).108. Wetzel, S. J. & Scherzer, M. Machine learning of explicit order parameters: From the Ising model to SU (2) lattice gauge theory. Phys. Rev. B96, 184410 (2017).109. Wetzel, S. J., Melko, R. G., Scott, J., Panju, M. & Ganesh, V. Discovering symmetry invariants and conserved quantities by interpreting siamese neural networks. Phys. Rev. Res.2, 033499 (2020).110. Friederich, P., Krenn, M., Tamblyn, I. & Aspuru-Guzik, A. Scientific intuition inspired by machine learning-generated hypotheses. Mach. Learn. Sci. Technol.2, 025027 (2021).111. Flam-Shepherd, D. et al. Learning interpretable representations of entanglement in quantum optics experiments using deep generative models. Nat. Mach. Intell.4, 544–554 (2022).112. Wellawatte, G. P., Seshadri, A. & White, A. D. Model agnostic generation of counterfactual explanations for molecules. Chem. Sci.13, 3697–3705 (2022).113. McGrath, T. et al. Acquisition of chess knowledge in AlphaZero. Preprint at arXiv 2111.09259 (2021).114. Schmidt, M. & Lipson, H. Distilling free-form natural laws from experimental data. Science324, 81–85 (2009).115. Gentile, A. A. et al. Learning models of quantum systems from experiments. Nat. Phys.17, 837–843 (2021).116. Cranmer, M. et al. Discovering symbolic models from deep learning with inductive biases. In 34th Conference on Neural Information Processing Systems (NeurIPS 2020) (NeurIPS, 2020).117. Georgescu, I. How machines could teach physicists new scientific concepts. Nat. Rev. Phys.https://doi.org/10.1038/s42254-022-00497-5 (2022).118. Cranmer, K., Brehmer, J. & Louppe, G. The frontier of simulation-based inference. Proc. Natl Acad. Sci. USA117, 30055–30062 (2020).119. Raayoni, G. et al. Generating conjectures on fundamental constants with the Ramanujan Machine. Nature590, 67–73 (2021).120. Wagner, A. Z. Constructions in combinatorics via neural networks. Preprint at arXiv 2104.14516 (2021).121. Rahwan, I. et al. Machine behaviour. Nature568, 477–486 (2019).122. Schmidhuber, J. in Workshop on Anticipatory Behavior in Adaptive Learning Systems, 48–76 (Springer, 2008).123. Pathak, D., Agrawal, P., Efros, A. A. & Darrell, T. in Proc. 34th International Conference on Machine Learning, 2778–2787 (PMLR, 2017).124. Thiede, L. A., Krenn, M., Nigam, A. & Aspuru-Guzik, A. Curiosity in exploring chemical spaces: intrinsic rewards for molecular reinforcement learning. Mach. Learn. Sci. Technol.3, 035008 (2022).125. Varshney, L. R., Rajani, N. F. & Socher, R. Explaining creative artifacts. Preprint at arXiv 2010.07126 (2020).126. Varshney, L. R. et al. A big data approach to computational creativity: The curious case of Chef Watson. IBM J. Res. Dev.63, 7:1–7:18 (2019).127. Itti, L. & Baldi, P. Bayesian surprise attracts human attention. Vision Res.49, 1295–1306 (2009).128. Schmid, U., Zeller, C., Besold, T., Tamaddoni-Nezhad, A. & Muggleton, S. in Proc. International Conference on Inductive Logic Programming, 52–67 (Springer, 2016).129. Muggleton, S. H., Schmid, U., Zeller, C., Tamaddoni-Nezhad, A. & Besold, T. Ultra-strong machine learning: comprehensibility of programs learned with ILP. Mach. Learn.107, 1119–1140 (2018).130. Feigenbaum, E. A. Some challenges and grand challenges for computational intelligence. J. ACM50, 32–40 (2003).131. de Regt, H. W. & Gijsbers, V. in Explaining Understanding: New Perspectives from Epistemology and Philosophy of Science (eds Grimm, S. R., Baumberger, C. & Ammon, S.) 50–75 (2017).