The 2021 AAAI Spring Symposia will take place online. Please stop back closer to the beginning of the symposium for the program. Please see the 2021 AAAI Spring Symposium site for registration information.
Blurb: In 2009, Jay Forrester (2009), the creator of system dynamics, remarked that we “live in a complex world of nested feedback loops,” caused by cascading interdependencies that exist between processes that vary in complexity and temporal scale. For computational systems employing AI/ML-based approaches that are using all available data about the functioning of such complex processes in order to estimate or learn various properties and states of the world or of a complex process, a Systems Engineering perspective can provide a framework for seeing interrelationships rather than things, for seeing patterns of change rather than static snapshots, as systems engineering methods are about seeing “wholes” (Walden et al., 2015). The increasing prevalence of AI/ML-enabled intelligent systems within our society will likely continue. This will reveal more nuanced interdependencies, in relation to the remarks made by Jay Forrester that AI-enabled intelligent systems have with both upstream processes that impact them, and downstream processes they affect. The benefits of studying these cascading interdependencies through a Systems Engineering perspective will be essential to understanding their behavior and further increasing the adoption of such complex system-of-systems in society.
Knowledge about current situations in the world is needed for a range of applications. Models of complex situations (“patterns of life”) are typically attacked with reductionism. However, in the absence of modeling dynamic and evolving feedback and interrelationships, these strategies fall short of expectations. If causal motivations for feedback and interrelations can be developed, it can be expected that the system modeling will be good, as well as the corresponding ability to estimate system dynamics. But real-world situations for Forrester’s “nested feedback loops” involve uncertainty and incompleteness, making accurate estimates difficult. These problems continue to be addressed by the Data and Information Fusion (DIF) communities, forming snapshots and estimates about component entities and states. However, the DIF community should seek to find opportunities where AI/ML can help move beyond component-level estimates toward modeling the synergies and dependencies within and across the feedback loops to achieve maximum situational awareness.
This Symposium will explore the cascading interdependencies of AI/ML-based software within the information processing pipeline through a Systems Engineering methodology (e.g., for interdependence in information systems, see Llinas, 2014). Cascading and interdependent phenomena affects the behavior of downstream components within a system. For instance, as new DIF capabilities emerge that can model synergistic effects across feedback loops (enabling greater situational awareness in reference to the JDL model), there will be consequences for downstream AI/ML-based systems that process the richer situational context to find patterns of significance for the human decision-maker (e.g., Snidaro et al., 2019). However, the decision-maker is also an effector in the complex systems, providing inputs and feedback into the AI-enabled system and can create further cascading interdependencies with all other internal downstream components. These relationships are further exacerbated by the unpredictability of the human decision-maker, uncertainty in the raw and fused data, and the large trade space of algorithm permutations that can be orchestrated to solve a given problem. Systems Engineering brings the opportunity to address and model the full range of complex, synergistic feedback loops in modern complex systems, toward the realization of cost-effective designs.
References:
Blasch, Erik, Cruise, Robert, Aved, Alexander J., Majumder, Uttam K. & Rovito, Todd V. (2019), Methods of AI for Multimodal Sensing and Action for Complex Situations, AI Magazine, 40(4), DOI: https://doi.org/10.1609/aimag.v40i4.4813
Blasch, E., Steinberg, A., Das, S., Llinas, J., Chong, C., Kessler, O., Waltz, E. & White, F. et al., (2013), "Revisiting the JDL model for information exploitation," Proceedings of the 16th International Conference on Information Fusion, Istanbul, pp. 129-136.
Forrester, Jay W. (2009), "Some Basic Concepts in System Dynamics," Sloan School of Management, MIT, D-4894, pp. 1-17; retrieved 8/23/2020 from https://www.cc.gatech.edu/classes/AY2018/cs8803cc_spring/research_papers/Forrester-SystemDynamics.pdf
Johnson, C. W. (2006). What are emergent properties and how do they affect the engineering of complex systems? Reliability Eng. Syst. Safety 91, 1475–1481. doi: 10.1016/j.ress.2006.01.008
Llinas, J. (2014). Reexamining Information Fusion–Decision Making Interdependencies, Presented at the IEEE CogSIMA conference, San Antonio, TX.
Ottino, J.M. (2004) Engineering complex systems. Nature 427: 399
Quadri, S. and Othman, S., (2012, Spring) Issue of Complexity in Data Fusion Systems, The Pacific Journal of Science and Technology, 13(1).
Sinha, Kaushik, and de Weck, Olivier L. (2013, August 4), Structural Complexity Quantification for Engineered Complex Systems and Implications on System Architecture and Design.” Volume 3A: 39th Design Automation Conference.
Snidaro, L., Garcia, J., Llinas, J. & Blasch, E. (2019), Recent trends in context exploitation for Information Fusion and AI, AI Magazine, 40(3): 14-27.
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. Hoboken, NJ: John Wiley & Sons.
In 2009, Jay Forrester (2009), the creator of system dynamics, remarked that we “live in a complex world of nested feedback loops,” caused by cascading interdependencies that exist between processes that vary in complexity and temporal scale. For computational systems employing AI/ML-based approaches that are using all available data about the functioning of such complex processes in order to estimate or learn various properties and states of the world or of a complex process, a Systems Engineering perspective can provide a framework for seeing interrelationships rather than things, for seeing patterns of change rather than static snapshots, as systems engineering methods are about seeing “wholes” (Walden et al., 2015). The increasing prevalence of AI/ML-enabled intelligent systems within our society will likely continue. This will reveal more nuanced interdependencies, in relation to the remarks made by Jay Forrester that AI-enabled intelligent systems have with both upstream processes that impact them, and downstream processes they affect. The benefits of studying these cascading interdependencies through a Systems Engineering perspective will be essential to understanding their behavior and further increasing the adoption of such complex system-of-systems in society.
Knowledge about current situations in the world is needed for a range of applications. Models of complex situations (“patterns of life”) are typically attacked with reductionism. However, in the absence of modeling dynamic and evolving feedback and interrelationships, these strategies fall short of expectations. If causal motivations for feedback and interrelations can be developed, it can be expected that the system modeling will be good, as well as the corresponding ability to estimate system dynamics. But real-world situations for Forrester’s “nested feedback loops” involve uncertainty and incompleteness, making accurate estimates difficult. These problems continue to be addressed by the Data and Information Fusion (DIF) communities, forming snapshots and estimates about component entities and states. However, the DIF community should seek to find opportunities where AI/ML can help move beyond component-level estimates toward modeling the synergies and dependencies within and across the feedback loops to achieve maximum situational awareness.
This Symposium will explore the cascading interdependencies of AI/ML-based software within the information processing pipeline through a Systems Engineering methodology (e.g., for interdependence in information systems, see Llinas, 2014). Cascading and interdependent phenomena affects the behavior of downstream components within a system. For instance, as new DIF capabilities emerge that can model synergistic effects across feedback loops (enabling greater situational awareness in reference to the JDL model), there will be consequences for downstream AI/ML-based systems that process the richer situational context to find patterns of significance for the human decision-maker (e.g., Snidaro et al., 2019). However, the decision-maker is also an effector in the complex systems, providing inputs and feedback into the AI-enabled system and can create further cascading interdependencies with all other internal downstream components. These relationships are further exacerbated by the unpredictability of the human decision-maker, uncertainty in the raw and fused data, and the large trade space of algorithm permutations that can be orchestrated to solve a given problem. Systems Engineering brings the opportunity to address and model the full range of complex, synergistic feedback loops in modern complex systems, toward the realization of cost-effective designs.
References:
Blasch, Erik, Cruise, Robert, Aved, Alexander J., Majumder, Uttam K. & Rovito, Todd V. (2019), Methods of AI for Multimodal Sensing and Action for Complex Situations, AI Magazine, 40(4), DOI: https://doi.org/10.1609/aimag.v40i4.4813
Blasch, E., Steinberg, A., Das, S., Llinas, J., Chong, C., Kessler, O., Waltz, E. & White, F. et al., (2013), "Revisiting the JDL model for information exploitation," Proceedings of the 16th International Conference on Information Fusion, Istanbul, pp. 129-136.
Forrester, Jay W. (2009), "Some Basic Concepts in System Dynamics," Sloan School of Management, MIT, D-4894, pp. 1-17; retrieved 8/23/2020 from https://www.cc.gatech.edu/classes/AY2018/cs8803cc_spring/research_papers/Forrester-SystemDynamics.pdf
Johnson, C. W. (2006). What are emergent properties and how do they affect the engineering of complex systems? Reliability Eng. Syst. Safety 91, 1475–1481. doi: 10.1016/j.ress.2006.01.008
Llinas, J. (2014). Reexamining Information Fusion–Decision Making Interdependencies, Presented at the IEEE CogSIMA conference, San Antonio, TX.
Ottino, J.M. (2004) Engineering complex systems. Nature 427: 399
Quadri, S. and Othman, S., (2012, Spring) Issue of Complexity in Data Fusion Systems, The Pacific Journal of Science and Technology, 13(1).
Sinha, Kaushik, and de Weck, Olivier L. (2013, August 4), Structural Complexity Quantification for Engineered Complex Systems and Implications on System Architecture and Design.” Volume 3A: 39th Design Automation Conference.
Snidaro, L., Garcia, J., Llinas, J. & Blasch, E. (2019), Recent trends in context exploitation for Information Fusion and AI, AI Magazine, 40(3): 14-27.
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. Hoboken, NJ: John Wiley & Sons.