An Entropy Approach to the Structure and Performance of Interdependent Autonomous Human Machine Teams and Systems (A-HMT-S)

Dear Colleagues, 

The development of autonomous teams and systems has become increasingly important. For our Special Topic, we are interested in developing the science of autonomy by advancing theory for human-machine teams and systems [1], its models, structure, performance and management of interdependent teams (e.g., with entropy or disorder); its central problems (e.g., control, machine learning, AI, the selection and mix of humans and machines); and its interdisciplinary issues (e.g., philosophy, ethics, trust, confidence, legal, social science). Possibly, computational swarms of independent agents increase the need for common goals or objectives, while interdependence increases the likelihood of system autonomy. This speculation leads us to suspect the existence of an overlap between the control of swarms and the governance of interdependent teams. If correct, autonomy presents a series of challenges that must be addressed. But we are left with more questions than answers (see our list at the end).

Theories based on traditional models using the independent data (i.e., i.i.d. models [2]) associated with Shannon [3] information theory have not been successful at replicating known interdependent effects (e.g., [4]), nor at predicting new effects (e.g., the inherent bistability of philosophical or political interpretations; of the shared tasks necessary to operate a restaurant business; of team competition in sports). Instead, social scientists in particular have been struggling to replicate their own findings [5], impeding them from being sufficiently confident to generalize to human-machine autonomy. In contrast, interdependence has been described as bewildering in the laboratory [6]. But it is a state-dependent phenomenon linked with similar phenomena [7]; e.g., quantum effects [8]. If we re-construe intelligence not as an individual quality that is to be prized, but, instead, as a more valuable phenomenon that occurs in the interactions of a whole team independent of the intelligence of its members [9], re-construal opens a path to theory and models that may advance the science of interaction among agents striving to form more productive social units composed of orthogonal parts (the complementary parts of whole teams, businesses, or systems). Moreover, state-dependent phenomena are dependent on multiple effects that contribute to a context in open systems, especially when having to make decisions while facing uncertainty ([10]; e.g., [11]), requiring a profound shift in what is considered observational. Although similar conclusions were drawn in social psychology [12] and Systems Engineering [13], Schrödinger ([14], p. 555) was the first to believe that a knowledge of the whole precluded full knowledge of the parts of a whole. Schrödinger attributed this loss of information at the individual level to the shift from independence to dependence. 

Independent models work well for closed system models. But closed models are often unrealistic. In a critique of closed systems, retired U.S. General Zinni [15] complained that the use of war games results in “preordained proofs”; that is, by choosing a game's context, a user can obtain a desired outcome. Moving from closed system models to open-systems, characterized by uncertainty, overwhelms traditional models; e.g., in economics, Rudd [16] began:

Mainstream economics is replete with ideas that “everyone knows” to be true, but that are actually arrant nonsense.

If the search for fitness is a search for dependency (viz., an adaptiveness that produces robustness), we seek models to advance autonomous human-machine systems, whether from economics; markets under threat (e.g., AT&T is shedding its media empire; in [17]); or disrupted relationships (divorces in marriage, business, science teams). In open systems, managing entropy production is critical.

Questions and Some Suggestions for Topics:

• We know that a team, composed of interdependent teammates, is more productive than the same members who work independently [18]; we do not know why, but we suspect offsetting entropy production from the complementary parts of a team when a highly interdependent team has been formed into a cohesive unit.
• Interdependence is state dependency. State-dependency models have achieved great predictive success in quantum mechanics while at the same time failing to be intuitive or to being open to a philosophical understanding [8];[19]. That highly predictive, state-dependent quantum models leave meaning open to interpretation makes models of interdependence non-traditional and non-rational, requiring a trial-and-error randomness in their structure [1]; how else are they identifiable other than by a system’s entropy production?
• It is common among philosophers to be pessimistic about a theory of meaning [20]. It may be, however, that a philosophical approach to the meaning of autonomy yields the very insight that autonomy researchers can use to build new theory.
• What does autonomy mean for humans or machines; swarms of machines; or society?
• How might autonomy differ for models of independent human-machine agents versus interdependent ones; the coevolution of humans and technology (p. 170 [21]); or organizational (e.g., the restaurant workers in dependent and orthogonal roles, such as a cook, waiter and clerk) and biological models of interdependent agents that perform in complementary and dependent team roles (e.g., biological collectives, like ants [22]; plants or “mother trees” [23])?
• A focus on independence is highlighted by the belief that “many hands make light work,” but this focus leaves the size of a team unsolved ([18], p. 33). Is the search for the fitness of a team’s or firm’s size the motivating cause of mergers or spin-offs? Does team size reflect the complexity of the problem being addressed?
• Can interdependence resolve the open-system questions posed by Rudd in economics and Gen. Zinni in games?
• From an interdisciplinary perspective, what might a non-traditional model of autonomous human-machine teams and systems look like?
• What metrics can be proposed for autonomous team structures and performance?
• How might the context of an autonomous team or system be determined when faced by the situations of uncertainty, conflict or competition that limit traditional models [24]?
• Can the governance of an A-HMT-S be conducted and explained with AI [25]?
• Should authority be given to a machine to take operational control from a human operator by overriding the human [26]? How much authority should be given to a machine in an interdependent team in the event its human operator becomes dysfunctional?
• If the governance of an A-HMT-S is designed to promote democracy, does that increase confidence in AI for autonomous human-machine teams and systems (e.g., [27], p.11)?

References:

1. Lawless, W.F. The interdependence of autonomous human-machine teams: The entropy of teams, but not individuals, advances science. Entropy 2019, 21, 1195. doi: 10.3390/e21121195.
2. Schölkopf, B. et al. (2021), Towards Causal Representation Learning, arXiv, retrieved 7/6/2021 from https://arxiv.org/pdf/2102.11107.pdf
3. Shannon, C.E. (1948), A Mathematical Theory of Communication, The Bell System Technical Journal, 27: 379–423, 623–656.
4. Google unit DeepMind tried—and failed—to win ai autonomy from parent (2021). https://www.wsj.com/articles/google-unit-deepmind-triedand-failedto-win-ai-autonomy-from-parent-11621592951
5. Nosek, B., corresponding author from OCS (2015), Estimating the reproducibility of psychological science, Science, 349 (6251): 943: also, https://doi.org/10.17226/25303.)
6. Jones, E. (1998), Major developments in five decades of social psychology,  In Gilbert, D.T., Fiske, S.T., & Lindzey, G., Handbook Social Psychology, Vol. I, pp. 3-57. Boston: McGraw-Hill.
7. Davies, P. Does new physics lurk inside living matter? Physics today 2020, 73, 34-41. doi: 10.1063/PT.3.4546.
8. Lawless, W.F. Quantum-like interdependence theory advances autonomous human–machine teams (A-HMTS). Entropy 2020, 22, 1227. doi: 10.3390/ e22111227.
9. Cooke, N. & Lawless, W.F. (2021), Cooke, N.J. & Lawless, W.F. (2021, forthcoming), Effective Human-Artificial Intelligence Teaming, In Lawless et al., Engineering Science and Artificial Intelligence, Springer.
10. Lawless, W.F.; Mittu, R.; Sofge, D.; Hiatt, L. Artificial intelligence, autonomy, and human-machine teams: Interdependence, context, and explainable ai. AI Magazine 2019, 40, 5-13. doi: 10.1609/ aimag.v40i3.2866
11. Colonial pipeline CEO Tells why he paid hackers a $4.4 million ransom (2021). https://www.wsj.com/articles/colonial-pipeline-ceo-tells-why-he-paid-hackers-a-4-4-million-ransom-11621435636?.
12. Lewin, K. (1951), Field theory of social science. Selected theoretical papers. Darwin Cartwright (Ed.). New York: Harper & Brothers.
13. Walden, D.D., Roedler, G.J., Forsberg, K.J., Hamelin, R.D. & Shortell, T.M. (Eds.) (2015), Systems Engineering Handbook. A guide for system life cycle processes and activities (4th Edition). Prepared by International Council on System Engineering (INCOSE-TP-2003-002-04. John Wiley.
14. Schrödinger, E., 1935. “Discussion of Probability Relations Between Separated Systems,” Proceedings of the Cambridge Philosophical Society, 31: 555–563; 32 (1936): 446–451.
15. Augier, M. & Barrett, S.F.X. (2021), “General Anthony Zinni (Ret.) on wargaming Iraq Millennium Challenge and competition," CIMSEC, https://cimsec.org/general-anthony-zinni-ret-on-wargaming-iraq-millennium-challenge-and-competition/
16. Rudd, J.B. (2021). Why Do We Think That Inflation Expectations Matter for Inflation? (And Should We?). Federal Reserve Board, D.C. https://doi.org/10.17016/FEDS.2021.062.
17. AT&T’s Hollywood Ending Erased Billions in Value; https://www.wsj.com/articles/att-hollywood-ending-erased-billions-value-hbo-discovery-warner-11621297279
18. National Research Council (NRC, 2015). Enhancing the effectiveness of team science. doi: 10.17226/19007.
19. Weinberg, S. (2017, 1/19), The Trouble with Quantum Mechanics, http://www.nybooks.com/articles/2017/01/19/trouble-with-quantum-mechanics; replies http://www.nybooks.com/articles/2017/04/06/steven-weinberg-puzzle-quantum-mechanics/
20. Speaks, J. (2021), "Theories of Meaning", Stanford Encycl. Philosophy, Edward N. Zalta (ed.), https://plato.stanford.edu/archives/spr2021/entries/meaning/
21. de León, M.S.P. et al. (2021), The primitive brain of early homo. Science, 372: 165-171. doi: 10.1126/science.aaz0032.
22. Insects (2021), Biological Collectives. https://ssr.seas.harvard.edu/insect-collectives.
23. Mother Tree Project (2021). https://mothertreeproject.org.
24. Mann, R.P. (2018), Collective decision making by rational individuals, PNAS, 115(44): E10387-E10396; from https://doi.org/10.1073/pnas.1811964115
25. Gunning, D et al., 2019, XAI—explainable artificial intelligence. Science Robotics, 4. doi: 10.1126/scirobotics.aay7120.
26. Sofge, D.A. et al. (2019), Will a Self-Authorizing AI-Based System Take Control from a Human Operator? From https://ojs.aaai.org/index.php/aimagazine/article/view/5196.
27. National Security Commission Report on AI (2021); https://www.nscai.gov/2021-final-report/

Dr. William Lawless
Dr. Donald Sofge
Dr. Daniel Lofaro
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Entropy is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.