Keywords

Introduction to Collective Intelligence

The term “intelligence” is used in this article to mean knowledge or information that informs action and as used in the sequence of understanding leading to wisdom: data, information, knowledge, intelligence, and wisdom (Tuomi 1999).

In the past, leaders would often gather wise elders and favorite consultants to discuss a problem until a solution was found. Then along came the Internet and Google, allowing leaders to have staff search through vast sources of information and distill these to provide intelligence for a decision. Meanwhile, the mathematically inclined might say, “give me your data and I will build a model to help you make the right decision.” All of these approaches have their value, so why not integrate them all into a system?

A collective intelligence system involves these three approaches, enabling each to improve the other in an ongoing feedback improvement system. I define collective intelligence as:

an emergent property from synergies among three elements: 1) data/info/knowledge; 2) software/hardware; and 3) experts and others with insight that continually learns from feedback to produce just-in-time knowledge for better decisions than any of these elements acting alone (Glenn 2009a)

Hence, collective intelligence can be thought of as continually emerging from changing synergies among its elements, as illustrated in Fig. 1.

Fig. 1
figure 1

Graphic illustration of three interactive elements of collective intelligence

Full size image

A useful and efficient collective intelligence system should connect these three elements into a single interoperable platform so that each of these elements can change the others. Participants should be able to comment on any information or software or computer model in the system. These comments read by reviewers and editors can then lead to changes in the system. For example, new insights from a person in an online group discussion can lead to changes or edits in the text of some part of the information in the system. This change of text might illustrate the need for new decision-making software or changes in one of the online computer models or the requirement to add a link to a new online computer model. Running the new model could produce results that, when given to the appropriate discussion group, could stimulate additional discussions leading to better insight that would lead to new edits in a situation chart. Decision support software like Delphi could add to the mix and result in changes in the information in the system and help identify new experts to add to the discussion groups. These changes in turn can lead to new questions in a Delphi, which in turn can change the text of information in the system.

Many of the features of a collective intelligence system (CIS) have existed before, but their integration into one platform creates a different experience, just as telephones and computers existed separately before email, but once integrated, the email experience was unique. The elements of the European opera existed before, but their integration into one experience is quite different than experiencing its different elements separately. A CIS can be thought of as a common platform of interlinked systems of people, information, and software each able to change due to feedback from the others.

Collective intelligence could be the next big thing in information technology (Glenn 2009a). CISs are relatively new developments within the ICT world, and hence, alternative approaches for creating them are more often described online than in professional journals (http://cci.mit.edu/, http://www.igi-global.com/journal/international-journal-organizational-collective-intelligence/1140, http://www-935.ibm.com/services/us/gbs/thoughtleadership/ibv-collective-intelligence.html, http://scripts.mit.edu/~cci/HCI/index.php?title=Main_Page, http://en.wikipedia.org/wiki/Collective_intelligence, http://shop.oreilly.com/product/9780596529321.do, http://www.dougengelbart.org/about/collective-iq.html). Nevertheless, there are already many different approaches to this subject. Here are a few. Pierre Levy of France focuses on the universal distribution of intelligence as the key element in Collective Intelligence: Mankind’s Emerging World in Cyberspace (Levey 1994). Anita Williams Woolley et al. explored the psychological factors that improve collective intelligence in groups in Evidence for a Collective Intelligence Factor in the Performance of Human Groups (Woolley et al. 2010). Francis Heylighen of Belgium has developed collective intelligence concepts to help the emergence of a “Global Brain” or “Global Brains” from the Internet (Heylighen 1999, 2008, 2013). He leads an international network to explore how to build a global brain at gbrain@listserv.vub.ac.be. MIT’s Center for Collective Intelligence has created the Handbook of Collective Intelligence as a wiki for an evolving conversation of the theory of collective intelligence (MIT Wiki). This center stresses the “peopleware” element of collective intelligence, looking at what characteristics are important in forming groups to best enhance their collective intelligence. The National Endowment for Science, Technology and the Arts in London has created a working draft discussion paper on collective intelligence as a Google Doc (Mulgan et al. 2011). Even the United Nations has been considered as a future center for global collective intelligence through a series of meetings and papers (Ekpe 2009).

Although one could consider Wikipedia, Google, crowdsourcing (Howe 2006), averaging expert judgments (Gordon 2009b), swarm intelligence (Kaiser et al. 2010), and prediction markets (Wolfers and Zitzewitz 2009) as examples of collective intelligence systems, these examples would not be a CIS by the definition offered in this article. They do produce information and in some cases, conclusions from a group, but they do not – so far – include feedback on a systematic, ongoing basis among the three elements to permit the continual emergence of new insights which then can affect other parts of their systems. They do not produce a continuous emergent intelligence, but only give a slice in time, whereas a CIS, like the mind, continually emerges and changes from the ongoing interaction of brain, experience, and environmental stimuli.

Some Historical Roots of Collective Intelligence Systems

Many efforts have been made to develop CIS over the years (Engelbart 2008). In the 1960s, Doug Engelbart at SRI created software and hardware to augment collaborative decision-making (Engelbart 1962). The Delphi method was developed at the RAND Corporation in the early 1960s and has subsequently been used by many organizations (Gordon 2009b). The SYNCON was developed in the early 1970s by The Committee for the Future which integrated discussion groups, video, and computer conferencing (Glenn 2009b). Murray Turoff’s pioneering Electronic Information Exchange System (EIES), also in the 1970s, paved the way for new thinking about collective intelligence (Hiltz and Turoff 1993) and in the author’s judgment provided the best example of a collective intelligence system at that time. The Wikipedia was created in 2001 (http://en.wikipedia.org/wiki/Wikipedia:About) and has grown to become the world’s most successful – if not the first truly global – participatory information and knowledge system with more than 76,000 active contributors working on over 31,000,000 articles in 285 languages as of August 2014. All of these make it seem that the emergence of Pierre Teilhard de Chardin’s Omega Point (de Chardin’s 1961) (integrated complexity and consciousness able to direct our evolution) and his popularization of Vladimir Vernadsky’s Noosphere seem inevitable.

Thomas Malone defined collective intelligence at the opening of the MIT Center for Collective Intelligence in 2006 as “groups of individuals doing things collectively that seem intelligent” (http://cci.mit.edu/about/MaloneLaunchRemarks.html). Subsequently, the Center’s website now lists its mission as: “How can people and computers be connected so that—collectively—they act more intelligently than any individuals, groups, or computers have ever done before (The MIT Center for Collective Intelligence) (Malone et al. 2010).” Thomas Malone and the MIT center continue to develop scholarly research on collective intelligence.

The Millennium Project has created collective intelligence systems for the Kuwait Oil Company (2003), the Climate Change Situation Room for Gimcheon, South Korea (2009), the Early Warning System for the Prime Minister’s Office of Kuwait (2010), its own Global Futures Integration System (2012) and is now creating E-ISIS (Egypt’s Integrated Synergetic Information System) for the Egyptian Academy of Scientific Research and Technology, which would be the first national CIS open to the public.

Collective Intelligence Systems Can Provide Support to Science and Technology Convergence

The accelerating complexity and the volume of change, with exponential increases in technological capacities and scientific knowledge, along with emerging interdependencies of economies, politics, and Internet-based groups, make it almost impossible for decision-makers and the public to gather and understand the information required to anticipate potential convergences among scientific knowledge and technologies to make and implement optimal or sufficiently robust decisions (Glenn 2008).

Because of such changes, the environmental, social, and security consequences of poorly informed decisions will have greater impacts tomorrow than they did yesterday. Hence, new systems for identification, analysis, assessment, feedback, and synthesis are urgently needed to inform decision-making. We need a more advanced system to think together about the future in some organized fashion so as to improve our collective prospects. We need a system to help us understand the global situation and prospects for the future that lets us interact with that information, discuss with colleagues, and use support software as need.

One approach to help identify potential future technological convergences is the combination of a modified cross-impact analysis (Gordon 2009c) and real-time Delphi (Gordon 2009a) within a CIS. For example, Table 1 below illustrates how experts could be invited to assess the future impact of one technology on another technology and the potential convergences.

Table 1 Technology convergence table
Full size table

Each cell (without the x) could be the subject of an ongoing Delphi, e.g., what are the impacts, convergent possibilities, etc. of nanotechnology on synthetic biology? As experts answer these questions, and feedback is given and responded to online, a collective intelligence would begin to emerge on the potentials of convergences and impacts. Information in each cell would be hyperlinked for better visual, user-friendly access.

One could also imagine the matrix above with hyperlinked submatrixes within each cell. For example, if there were five major approaches to nanotechnology, then a cross-impact matrix of five by five could be hyperlinked in the first cell with the X. One could further imagine a sub-five by five matrixes in Cell “A” inviting participants to cross-impact five nanotechnology approaches with five synthetic biology approaches. This could be a complex, but visually managed by online hyperlinking. In this way, the general public could see – and potentially comment on – the main impacts, and the more advance scientists and engineers could hyperlink into greater detail in the sub-cells of the matrix. As a result, vast and complex information would be available in an organized, user- friendly way for both the public and the professional.

We lack systems that make it easy to see and update a situation as a whole, including current conditions and forecasts, desired situation with a range of views, and alternative policies to address the gap between what is and what ought to be. Instead, analysts often use the Internet to go from one source to other, becoming stressed by information overload (Blair 2013).

It is wise to get all the relevant information before assessing potential future S&T integration or making any informed decision, but it is increasingly difficult to organize all the positions, priorities, and strategies, in a way that brings more satisfactory coherence to our thinking and decision-making. Or as Leon Fuerth put it in Anticipatory Governance:

There are many sources of foresight available to decisionmakers originating both within and outside of the U.S. Government, but foresight is not methodical, continuous, or structured in a form that is useful for decisionmakers… A simple collective intelligence system (CIS) would manage content, organize expertise, track comments and changes in documents, and support prioritization. (Fuerth and Faber 2012, pp 19 and 51)

Since everyone faces similar problems of managing complexity and change, we can expect to see the increased creation and customization of collective intelligence systems by governments, corporations, NGOs, universities, associations, and individuals. It has also been suggested that a collective intelligence system could provide continuity from one government administration to the next, “by making it easier to retain and transfer institutional knowledge that is essential for long-term strategic coherence, regardless of changes in policy or political philosophy (Fuerth and Faber 2012, pp 51).”

Collective intelligence systems can also focus on specific issues such as climate change or industries like agribusiness. Just as spreadsheets have become a general tool that any organization or individual can use, so too, CIS could become a general tool adapted for an individual or organization, or a country, as in the case of Egypt to help “bring it all together” (The Millennium Project).

An Example of a Collective Intelligence System Used to Support Global Future Research

The Global Futures Intelligence System (GFIS) was created by The Millennium Project and launched at the Woodrow Wilson International Center for Scholars in Washington, D.C., in January 2013 (Video of the launch of GFIS). It is available at www.themp.org (Fig. 2).

Fig. 2
figure 2

Front page of the Global Futures Intelligence System at www.themp.org

Full size image

Using the GFIS platform alone, one can track and anticipate global change within the framework of 15 Global Challenges, review specific global futures research on a variety of topics, and access internationally peer-reviewed chapters on 37 futures research methods. Questions can be addressed and discussed with experts on the challenges, research, and methods. For example, if one wanted to explore the convergence of computational science and the Internet of things, one could discuss which methods to use, identify potential participants in the database, and send invitations to each participant. If a real-time Delphi is chosen, then it can be collaboratively designed, managed, results analyzed, and report finalized for downloading, all on one integrated platform.

GFIS is an example of the next generation of interactive technology to support those exploring global change and potential futures for humanity. It began by integrating all of The Millennium Project’s 10,000+ pages of futures research obtained over the past 16 years from the annual State of the Future reports, plus information from expert listserv groups and its 50 Nodes (groups of individuals and institutions that connect local and global research and perspectives) around the world. It also includes the Futures Research Methodology version 3.0 with 39 chapters totaling more than 1,300 pages of internationally peer-reviewed methods for exploring the future (37 chapters on specific methods plus an introductory chapter and a concluding chapter). All of this, plus its software, are integrated in one platform so that users can participate to update and improve any element of this online collective intelligence system.

The most commonly used section is the 15 Global Challenges that helps see the potential convergence of science and technology but also its relationships and integration of other areas (Table 2).

Table 2 Menu options for each of the 15 Global Challenges
Full size table

One of the greatest values of GFIS is that it serves as an interactive decision support dashboard, rather than just another source of information. It offers more than just new software tools – a vast body of intelligence/knowledge/data and access to experts; it is also an evolving system for producing synergies among the three elements of collective intelligence (people, information, and software) that have greater value than the sum of their separate, individual values.

Figure 3 below shows an example of a “situation chart” for organizing information to reflect the present and desirable situations related to a specific challenge, as well as polices to address the gaps between what is and what ought to be. As with other elements of GFIS, the situation charts are continuously updated based on feedback and new information gathered throughout GFIS.

Fig. 3
figure 3

Example of a situation chart: Global Challenge 1. How can sustainable development be achieved for all while addressing global climate change?

Full size image

Everything in GFIS can be commented on by anyone who accesses the system. Substantive edits, updates, and improvements can be made in real time by experts and the public via a rapid review process. GFIS reviewers will be notified automatically of any suggested edits within their expertise. Since reviewers are very busy, not expert in every detail of a challenge, and might not have seen the suggestion, it is assumed that of the 100 reviewers per challenge, at least three will give their review of the suggestion to the relevant challenge within 24 h. These reviews go to the GFIS editors who then make the edit (Table 3).

Table 3 The types of participants, their roles, and how they are selected
Full size table

Examples of How One Might Use GFIS to Better Understand and Anticipate Science and Technology Convergence to Benefit Society

  • Select “Updates” of the whole system from the front page. This gives all recent changes in scanning items, edits to reports and situation charts, RSS news aggregator, comments, discussions, questions, web resources, books, papers, and models. The results are displayed in an executive dashboard allowing one to go into further detail on any category and any item in a category. This can help one to see how their interest fits into the larger whole. A specific term can be entered that returns the most recent matches.

  • Pose questions. Results from the search can lead to new questions that can be posed in the relevant discussion groups or added as a structured question in a real-time Delphi setting-specific demographic categories for analyzing the results. The categories could be from nonscientist or non-engineer as well to get a broader range of views helping to anticipate potential convergences.

  • Conduct searches in specific challenges that might not seem related to a specific S&T item. One could search within the global challenge on organized crime and find connections to S&T not previously considered. One can search by expert reviewers’ inputs versus nonexpert inputs to the system. These searches might bring up an important question that can be submitted to the discussion areas to obtain broader feedback and new insights from the group to the relevant challenge group. The discussion might generate several suggestions for updating or improving the text of the challenge’s short overview. These suggestions are automatically sent to the reviewers of the challenge and then, if accepted and/or amended, are entered into the regular text of the challenge.

  • Create a technology converge table to help anticipate S&T converges via integration of cross-impact analysis and Delphi. One example is explained above in Table 1. Technology convergence table gives another way to gain new insights that might not have been known before.

GFIS is constantly evolving. Its collective intelligence will increase with the number of users, review and improvement of information, new software and models, and better integration of all these elements of a collective intelligence system.

Conclusion

CIS can be a new tool for both the advanced scientist and the general public to help anticipate S&T convergence to benefit society. Expert judgment and public option can be organized to increase our collective intelligence about increasingly complex issues on one platform. Agreements about S&T policy can be identified and strategies developed in an open, transparent fashion. This is also a potential new source to improve media coverage of S&T issues to make our public dialogues more informed that the current fragmented information systems.