Although it is likely that human-computer decision-making systems will continue to advance, a complete path forward is not yet clear. As illustrated in this report, important progress is being made in a number of underlying technologies and scientific foundations, and in particular there is a good deal of innovation in human-computing interfaces. However, the committee identified three general challenges:
1. Scientists do not fully understand the human decision-making process. That understanding is being built up in multiple fields, such as cognitive science, cultural anthropology, decision science, neuroscience, and psychology. Without a more complete understanding of how humans decide, it is difficult to know how far advances in human-machine decision making can go. We do not know all the enablers of good decisions and how those enablers might be turned against us. What is the likely progress for those enablers over the next 20 years, and what are the metrics to track in order to discern progress? Are others likely to move ahead of the U.S. on any of these enablers? How can we integrate all these enablers in order to improve data-to-decisions? All of these fundamental questions require further investigation.
2. There is no “silver bullet.” Enhancing human-machine collaboration does not solely depend on finding the right algorithms, or on improving computerized language processing, or on designing a more natural interface between humans and machines, or on resolving challenges associated with “big data” and so forth. Rather, all of these solutions and more are needed. Indeed, although this report touches on 11 different fields and subfields,1 these represent just some of the scientific approaches that could be included to enhance human-machine collaboration for decision making. While the problem is profoundly multidisciplinary, university departments—both in the United States and elsewhere—are still largely focused on individual fields. Even in those universities where exciting multidisciplinary research is conducted there are limits to how far researchers tend to go outside their own subject matter, such as learning, critiquing, and adopting one another’s terminology and concepts. It is possible that public or public-private institutions, such as the Agency for Science Technology and Research and the German Research Centre for Artificial Intelligence (described in Appendix B), may offer innovative approaches to interdisciplinary research.
3. There is a need to better understand the social implications of human-machine collaboration for decision making.2 Whether machines ought to “decide” when to pull the trigger has been
1 Artificial intelligence, cognitive science, computer science, data analytics, decision science, machine learning, natural language processing, neuroscience, psychology, statistics, systems engineering.
2 A useful discussion of these issues may be found in Emerging and Readily Available Technologies and National Security—A Framework for Addressing Ethical, Lethal and Societal Issues. National Academies Press, 2014.
discussed broadly.3 But as machines become better decision makers, will humans increasingly defer to them? Should they? What will happen to human cognitive processes as humans gain greater fluency with computing, especially through early childhood formal and informal learning? How should the need for privacy (by the government as well as the individual) be assessed relative to the ability to fully harness the potential benefits of data sharing?
The committee identified a number of promising research directions to improve the scientific basis for strong human-computer decision making and to help inform these open challenges:
- Data-to-decisions is an umbrella term that is not clearly defined. We need a better understanding of how cognitive functions can be supported over time and in context and an overall framework for thinking about how to design human-computer decision systems; The ubiquitous capability to capture, store, reproduce, move, and reuse data has led to decisions increasingly being made by networks composed of humans and machines. Yet, the exploitation of that data is often ad hoc. Research is needed to frame and systematize how we exploit that data;
- At any moment, whether a particular datum will be relevant or irrelevant into the future is task and context dependent, so there is an incentive to retain more, rather than less. Thus, a key challenge is to build task and context models that enable data to be filtered and processed into “useful information”;
- Another challenge is developing systems that allow both humans and computers to work together in a harmonious team, rather than one supervising the other. This requires research to help individual and team exploration of (partial and incomplete) hypotheses, to enable continuous learning by the system (e.g., so the system can learning how to predict an analyst’s needs and preferences, to guide continuous ingesting of data and its metadata and fusing it into the existing data, to cue decision makers to relevant, unexplored data or behavior; and to facilitate the sharing of hypotheses and derived knowledge among team members (such as by developing languages that make it easy for decision makers to state what they want the data to tell them). Creating harmonious human-computer teams would also be helped by research in comparing the different roles of humans and computers in mixed teams;
- Complex decision making often takes place in a complex environment, with multiple activities occurring simultaneously. This leads to frequent interruptions and the need to switch tasks and revise priorities. Current human-computer systems do not handle interruptions well and they need to provide more support for the resumption of interrupted activities. More research is needed on computational interruption management techniques and algorithms, rooted in an understanding of people’s cognitive and attentional capabilities; and
- More work is needed to develop a methodology for evaluating and assigning metrics for each individual piece of the collaboration and for the quality of the decisions made by the overall human-machine collaborative system.
3 The numerous articles about the use of drones in military combat are just one example.
The committee members found more questions than answers during the course of this study. Their observations, however, do not call into doubt the importance of future human-machine collaboration for complex decision making as much as they underscore a present-day reality: The development of human-machine collaboration for complex decision making is still in its infancy relative to where cross-disciplinary research could take it over the next generation.