Skip to article
AHFE AHFE Discoveries AHFE International Open Access Journal
AHFE OPEN ACCESS OPEN ACCESS CONFERENCE PROCEEDINGS PAPER

Workshop: Orchestrating Synthesized Human and AI-Agentic Workflows: AI Agency Benefits, Disruptions and Management

Publication year
Digital Object Identifier10.54941/ahfe1007672

Abstract

For our Special Interest Group (S.I.G.), we propose papers that address the orchestration of teams by synthesizing their workflows into a coherent whole, whether these teams are composed of human, machine, Generative AI (gen-AI), robot or AI-Agentic members. The bigger picture of interdependence, teamwork and Gen-AI indicates the need by organizations to build a library of human and artificial agents with bidirectional agency (responsibility) to achieve operational goals (missions), considering agentic risk tolerances, available skills, and vulnerabilities across a complex trade space among the skills available versus those needed for the tasks assigned to complete an operation. In this trade space, agents (human or artificial) from multiple systems with the requisite skills to accomplish a designated task and timeline combined to form a hierarchy of humans, robots, machines and AI. This complex system produces workflows that must be synthesized into a unit(s), then orchestrated to accomplish the goals assigned to it, yet remain trusted even in competitive and uncertain environments. Once synthesized into a unit (e.g., a team), Gen-AI provides the opportunity to not only advance the science of teams by orchestrating team products and performances, but also has raised several concerns (viz., AI used for deception, superintelligence, blackmail, or existential threats to humans). For our S.I.G., We are interested in orchestrating teams: What are the benefits, drawbacks, and, most importantly, can humans, machines and Agentic AI be synthesized and managed (orchestrated)?

INTRODUCTION

The bigger picture of interdependence, teamwork and Gen-AI under uncertainty indicates the need by teams, organizations and institutiuons to build a library of human and artificial agents with bidirectional agency (viz., shared responsibility, individual and team accountability, mutual trust, etc.) to achieve operational goals (missions), considering agentic risk tolerances, available skills, threats and vulnerabilities across a complex trade space among the skills available versus those needed for the tasks assigned to complete for a team its operations and set of tasks, without conflicting with the operations

of a larger organization. In this trade space, agents (human or artificial) from multiple systems with the requisite skills to accomplish a designated task and timeline combine to form a hierarchy of humans, robots, machines and AI. This complex system produces workflows that must be synthesized into a unit(s), then orchestrated to accomplish the goals and tasks assigned to it, yet remain safe to users and trusted even in competitive, difficult and uncertain environments (reviewed in Lawless & Moskowitz, 2025).

Our interests are in exploring the architecture of structures, strategies and concerns for synthesizing effective and efficient engineered teams and systems employing both Gen-AI processes, human and agentic intelligence for a wide range of applications. The knowledge for interdependent dynamics across such disparate processes has yet to be sufficiently understood to achieve effective and emotional balance in real-world applications, especially for autonomous systems requiring predictable and highly reliable performance with complex tasks (Lawless & Moskowitz, 2025).

Large language model (LLM) agents can be constructed to undertake tasks in agent systems with code that orchestrates LLM calls and provides the tools needed to adapt as tasks in the field evolve, sometimes rapidly or on the fly as in the war in Ukraine. But as a possible issue for syntheses, an increasing number of Gen-AI systems can modify their own source code to improve system performance and to be able to rapidly adapt to new tasks in the field (Robeyns et al., 2025).

There are reports of advantages from the use of Gen-AI. For example, from The Atlantic (Wong, 2026), Gen-AI has shown the ability “to push the frontier of human knowledge and civilization.” But whether AI is learning, understanding, or just plain memorizing its responses remains an open question that needs to be addressed (e.g., Reisner, 2026).

Further, a report by the National Academy of Sciences (Burley et al., 2025) emphasized the value of interdependence, cooperation, trust, and engineering the size of a team to fit a targeted problem. Yet it seemed unable to address how team fit would occur, and unable to generalize the latter at this time, thereby highlighting the limits of classical team science (Lawless, 2025), previously predicted by theory (Lawless et al., 2023), and even supported by the National Academy’s earlier report with its claim about the loss of information during team interactions: the “performance of a team is not decomposable to, or an aggregation of, individual performances” (Endsley et al., 2022, p. 12).

A different National Academy (NASEM, 2025) panel addressed AI-risk mitigation in future NASA missions to the Moon and Mars; the benefits of supporting collaboration; using AI to build trust; and understanding the risks of using AI and managing its risks, while not addressing the advantages gained by synthesizing Gen-AI, machines, and humans into teams that can be orchestrated. From Kaplowitz’s (2025) discussion for AI, machines (viz, drones) and humans involved in modern military operations, Ukraine’s “operation spider web” against Russia, and Israel’s “successful military operations” against Iran show the value of synthesizing AI, machines and deception into forming an operational prowess that can be orchestrated, placing organizations at a disadvantage that do not or cannot synthesize

Orchestrating Synthesized Human and AI-Agentic Workflows 3 nor orchestrate advanced systems (for a review of AI already in use by the military, see Zequeira, 2026).

But in addition to a synthesis and an orchestration of the result, there are cautions, too. Aligned, engineered models are believed to prevent abuses by Gen-AI; however, even after becoming well-trained, Gen-AI aligned models are susceptible to being easily misaligned, manipulated and corrupted (Betley et al., 2025). From the Wall Street Journal (Berg & Rosenblatt, 2025), “Twenty minutes and $10 of credits on OpenAI’s developer platform exposed that disturbing tendencies lie beneath its flagship model’s safety training.”

Expertise: Our research teams have participated in 2 prior AHFE Special Interest Groups (S.I.G.):

1. 2025: “Generative AI Risks and Benefits within Human-Machine Teams,” Sessjons 110 and 125, July 30th, leading to a book: Lawless, Mittu, Sofge & Brambilla, 2025, Bidirectionality in Human-AI Collaborative Systems, by Elsevier; and, 2. 2024: “Data dependency,” Sessions 43, 69, 186, and 211, July 22–24, 2023, leading to a book: Lawless, W.F., Mittu, R., Sofge, D.A., & Fouad, H. (Eds.), (2024), Interdependent Human-Machine Teams: The path to autonomy. Elsevier.

A Non-Exhaustive List of Topics on AI Agency Below, we provide a suggestive but non-exhaustive list of possible topics for AHFE-2026:

Topic 1: How do we know that an operation’s timeline, agency, and task assignments are realistic? Do we know the indicators of success and failure, the engineering of synthesis, the fitness of agents, the welfare needed for each member, and the engineering “how” of orchestrating a team? Going deeper, how do we characterize the mutual fit among agent skills and tasks? Can engineered structures of teams produce optimal task completions? Or does more research need to be conducted to better understand the needs of how to manage autonomous systems performing the tasks that have been chosen for them to be performed while being viewed by the public?

Topic 2: To orchestrate, what information needs to be communicated or training perfored about trust uncertainty to remain a trustworthy team or teammate? How can AI observe teammates collaborating across different workstations and predict where visual attention for each goes next? Could the system decipher where visual attention is lacking and guide it for teammates accordingly (e.g., similar to autonomous lane management in new cars; see Heaps, 2025)? What are the potential pitfalls in this role for AI, Gen-AI or AI Agents? (For more, see the MIT Technology Review, Heaven, 2026).

Topic 3: To synthesize Gen-AI and humans, can we predict and control the benefits (e.g., reaching consensus, in Tessler et al., 2024), bidirectional trust issues, agentic resilience, as well as the potential impediments such as hallucinations (e.g., made-up results) and existential threats (e.g., runaway code; in Bengio et al., 2025a; scams against unsuspecting users; in Shojaee et al., n.d.) for users, bystanders and society, especially from superintelligent

agents (Bengio et al., 2025b; see also Shojaee et al., 2025; and Schroeder et al., 2026)?

Topic 4: During an orchestration, could AI be used to monitor teams and parse out the information that is being gleaned as trust is being built? What information, complete or not, is produced during interactions (viz., the independent and indentically distributed, or i.i.d., data which by definition is unable to reconstruct an observed interaction; in Schölkopf et al., 2021)? How is bidirectional trust expressed between humans and machines (e.g., what sensor information leads an F-16 fighter jet to take control from its fighter- jet pilot; in Centeno, 2024)? What similar actions of control are taken by ships to avoid collisions (viz., most ship colllisions are caused by human error, which AI can help to prevent; in Yoshioka & Hashimoto, 2025)? What are the cybersecurity issues for operational autonomous cars (Bhemavarapu, 2025)?

CONCLUSION

Human-machine-AI-robot teams are becoming more and more likely. But these teams make for complex systems that create even more complex workflows, yet must be performed safely. Each of these systems must be synthesized into a unit that is able to not only accomplish the goals set for it, but they must be orchestrated to remain trusted even in competitive, treacherous and uncertain environments. Once synthesized into a unit (e.g., a team, an organization, an institution), Gen-AI provides the opportunity to not only advance the science of teams by orchestrating team products and performances while remaining trusted by users and by the public, but also it has raised several concerns (viz., AI used for deception, superintelligence, blackmail, or existential threats to humans). Each of these concerns are valid and must be addressed satisfactorily to keep the trust of users, organizational owners, and their public.

ACKNOWLEDGMENT

The authors would like to acknowledge the lively discussions that preceeded the development of the AAAI-26 workshop that has also led to this Special Interest Group session. In addition, the first author would like to thank Ranjeev Mittu for his many years of support and leadership; and thanks also to Ira Moskowitz, for his mathematical guidance.

REFERENCES

Bengio et al. (2025a). International AI Safety Report. The International Scientific Report on the Safety of Advanced AI. DSIT 2025/001. Retrieved 2/22/2025 from https://assets.publishing.service.gov.uk/media/679a0c48a77d250007d313ee/ International_AI_Safety_Report_2025_accessible_f.pdf Bengio, et al. (2025b), Superintelligent Agents Pose Catastrophic Risks: Can Scientist

AI Offer a Safer Path? arXiv, arXiv:2502.15657 Berg, C.; Rosenblatt, J. (2025, 6/26), “The Monster Inside ChatGPT. We discovered how easily a model’s safety training falls off, and below that mask is a lot of darkness,” Wall Street Journal, retrieved 7/7/2025 from 796ac9d3. Betley et al. (2025), Emergent Misalignment: Narrow finetuning can produce broadly

misaligned LLMs, arXiv, https://arxiv.org/abs/2502.17424, 2025.

Orchestrating Synthesized Human and AI-Agentic Workflows 5 Bhemavarapu, S.V.R. (2025), Cybersecurity for Autonomous Vehicles, arXiv:

2504.20180. Burley, D. et al. (2025), The science and practice of team science. National Academies of Science, Engineering and Medicine. Washington DC. Centeno, G. (2024), F-16: how the fighter has evolved 50 years after its first flight,

Aeroflap, retrieved 10/10/2024 from https://www.aeroflap.com.br/en/f-16-how- the-fighter-evolved-50-years-after-its-first-flight/. Endsley, M.R., et al. (2022). Human-AI Teaming: State-of-the-Art and Research Needs. The National Academies of Sciences-Engineering-Medicine. Washington, DC: National Academies Press, https://www.nap.edu/catalog/26355/human-ai- teaming-state-of-the-art-and-research-needs Heaps, R. (2025, 5/29), Lane-Keeping Assist: Everything You Need to Know, Kelley Blue

Book, retrieved 2/25/2026 from https://www.kbb.com/car-advice/lane-keeping-assist Heaven, W.D. (2026, 1/12), “Meet the new biologists treating LLMs like aliens. By studying large language models as if they were living things instead of computer programs, scientists are discovering some of their secrets for the first time,” MIT Technology Review, retrieved 1/15/2026 from https://www.technologyreview. com/2026/01/12/1129782/ai-large-language-models-biology-alien-autopsy/ Kaplowitz, A. (2025, 8/3), “Can AI and Drones Replace Soldiers and Jets? Strikes

by Ukraine and Israel show modern militaries need to blend military tech old and new,” Wall Street Journal, from 7c66e9dc. Lawless, W.F. & Moskowitz, I.S. (2025), Editorial, An Entropy Approach to Interdependent Human-Machine Teams. Entropy, 27(2):176, doi: 10.3390/ e27020176 Lawless, W.F. (2025), Interdependent Human-Machine Teams: The Limits of Classical

Team Science, Proceedings of the 2025 AAAI Spring Symposium Series, 5(1): 87–89,. DOI: https://doi.org/10.1609/aaaiss.v5i1.35564 Lawless, W.F., Mittu, R., Sofge, D.A., & Fouad, H. (Eds.), (2024), Interdependent Human-Machine Teams: The path to autonomy. Elsevier. Lawless, W.F., Moskowitz, I.S.; Doctor, K.Z. (2023), A Quantum-like Model of

Interdependence for Embodied Human–Machine Teams: Reviewing the Path to Autonomy Facing Complexity and Uncertainty, Entropy, 25, 1323. https://doi. org/10.3390/e25091323 Lawless, W.F.; Mittu, R.; McGarry, S.P.D.; & Brambilla, M. (underway; forthcoming 2026). Generative AI Risks and Benefits within Human-Machine Teams. Elsevier. Lawless, W.F.; Mittu, R.; Sofge, D.A.; & Brambilla, M. (2025). Bidirectionality in

Human-AI Collaborative Systems. Elsevier. National Academies of Sciences, Engineering, and Medicine (NASEM). 2025.

Human and Organizational Factors in AI Risk Management: Proceedings of a Workshop. Washington, DC: The National Academies Press, pp. 29–30, https://doi. org/10.17226/29046. Reisner, A. (2026, 1/9), “AI’S MEMORIZATION CRISIS. Large language models don’t “learn”—they copy. And that could change everything for the tech industry,” The Atlantic, retrieved 1/16/2026 from https://www.theatlantic.com/ technology/2026/01/ai-memorization-research/685552/ Robeyns, M.; Szummer, M.; Aitchison, L. (2025), A Self-Improving Coding Agent,

arXiv: 2504.15228. Schölkopf, B., Locatello, F., Bauer, S., Ke, N.R., Kalchbrenner, N., Goyal, A. & Bengio, Y. (2021), Towards Causal Representation Learning, arXiv, retrieved 7/6/2021 from https://arxiv.org/pdf/2102.11107.pdf Shojaee, P. et al. (n.d.), Apple, Inc., https://ml-site.cdn-apple.com/papers/the-illusion-

of-thinking.pdf ;

Shojaee, P., Mirzadeh, I., Alizadeh, K., Horton, M., Bengio, S., and Farajtabar, M.

(2025), The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity, Machine Learning Research, accssed July 15, 2025 from https://machinelearning.apple.com/research/ illusion-of-thinking Schroeder, D.T. et al. (2026), How malicious AI swarms can threaten democracy. The fusion of agentic AI and LLMs marks a new frontier in informnation warfare, Science Policy Forum, 391(6783): 354–357, DOI: 10.1126/science.adz1697 Tessler et al. (2024), AI can help humans find common ground in democratic

deliberation, SCIENCE, 386 (6719): DOI: 10.1126/science.adq2852 Wong, M. (2026, 2/25), “The Edge of Mathematics. Terence Tao, the legendary mathematician, explains the promise of generative AI,” The Atlantic, retrieved 2/25/2026 from https://www.theatlantic.com/technology/2026/02/ai-math-terrance- tao/686107/ Yoshioka, H. & Hashimoto, H. (2025), Explainable AI for ship collision avoidance:

Decoding decision-making processes and behavioral intentions, Applied Ocean Research, 156, 104471, ISSN 0141-1187, https://doi.org/10.1016/j.apor.2025.104471. Zequeira, M., Major, U.S. Army, Artificial Intelligence as a Combat Multiplier. Using AI to Unburden Army Staffs, retrieved 2/25/2026 from https://www.armyupress. army.mil/Journals/Military-Review/Online-Exclusive/2024-OLE/AI-Combat- Multiplier/

Cover of the proceedings volume Human Factors in Robots, Drones and Unmanned Systems

Explore the complete digital volume

Human Factors in Robots, Drones and Unmanned Systems

Read this individual research paper free through the original publisher, or explore the full digital proceedings book.

ISBN 978-1-964867-94-6

Explore this ISBN in the AHFE Bookstore Listed in the July 2026 bookstore catalogue; search by ISBN to confirm current availability · Digital book purchase optional

Explore Related Research

Related research 8 matches

More papers from this volume

← Back to Research ExplorerBrowse the full library →