The AAAI 2023 Spring Symposium on LifeCycle Transparency


Topic:

Transparency is the ability for the operator to know, or see—as ‘through glass’—the workings of a machine or automation. Increasing transparency generally improves human situation awareness, trust, and human-machine performance. To have these benefits, the human must absorb the “transparency information” and use it to first understand what automation is doing and then determine whether and how to intervene or support it. This can require processing yet more information precisely when the operator is least able to do so.

LifeCycle Transparency(LCT) (Miller, 2021) looks to reduce this additional cognitive load by shifting it outside the execution window. LCT tries to increases the amount of transparency information transferred in pre- and post-mission phases to help ensure that the human operator understands (and can affect) how the automation performs at execution time. Note, pre-mission planning, post-mission debriefs and explanations, and even initial design can serve this role, but only if they inculcate accurate understanding of the automation's behavior. This symposium will discuss, clarify, and formalize LifeCycle Transparency, as well as to define paths forward for adopting a life cycle perspective into automation design, deployment and use.

Format:

The symposium will include paper presentations, invited talks, and panel discussions. Expressions of interest in participating in the panels are welcome at cmiller@sift.net.

Keynote Speakers:

Our Keynote speakers will include:

and we will be posting the titles and abstracts of their talks here as they become available.

Submission Guidelines:

The symposium will be accepting submissions in three forms with differing lengths, differing review processes. To encourage participation and focus on work in progress, we will consider submissions of previously published papers or papers that are under review to other venues. Work that has been previously published or is under submission should be noted in the title.

Submission types:

ABSTRACTS/POSITION PAPERS: For all individuals planning on attending the symposium we request they submit an abstract/position/experience statement of up to one(1) page in length (including references) about their views on the issues facing LCT, its technical and formal requirements and needs.

SHORT PAPERS: Short papers are intended to present some small result or idea. As such, short paper submissions may be up to three(3) pages including references. Their acceptance for presentation will be at the discretion of the symposium chairs and would not be considered peer reviewed. Such submissions that are accepted will have the opportunity to provide a longer reviewed paper for inclusion in the symposium's proceedings.

FULL PAPERS: Full paper submissions may have up to eight(8) pages including references. Full papers will be reviews by at least three members of the organizing committee. Those that are accepted will be presented at the symposium and will have the opportunity to be included in the symposia's proceedings.

Submission Format:

All submissions must be formatted in AAAI two-column, camera-ready format. Papers must be in trouble-free, high-resolution PDF format, formatted for US Letter (8.5" x 11") paper, using Type 1 or TrueType fonts. We refer authors to the 2023 AAAI conference author kit for details.

Submission website and deadline:

The symposium will use easychair for all submissions. The deadline for all submissions is February 17th, 2023 anywhere in the world.

Co-Chairs:

Tentative Schedule

Day 1

Day 2

Day 3 - TBA

LifeCycle Transparency Background

Transparency in automation and AI systems is the ability for the operator to know, or see—as ‘through glass’—the workings of the machine. Bhaskara, Skinner and Loft’s recent (2020) review of the literature on transparency says that “In a transparent system, information regarding the agent’s actions, decisions, behavior, and intentions is communicated to the operator through an appropriate interface with the aim of improving trust in the system, performance, and operator situation awareness (SA).” Increasing the transparency of a system has generally been shown to improve human SA, trust and, frequently, overall human-machine performance compared to systems which include less or no transparency (Mercado, et al., 2016; Lyons & Havig, 2014; Ososky, Sanders, Jentsch, Hancock, & Chen, 2014; de Visser, Cohen, Freedy, & Parasuraman, 2014).

How we commonly think about, report and research the concept of transparency today presents a problem, however. Automation is generally installed precisely because human operators do not have time, skill, attentional capacity or adequate precision to perform the task that the automation does and/or because those human resources must be devoted to some other task. Yet, for transparency to function and yield benefit, the human must be able to absorb the “transparency information” and use it to first understand what the automation is doing and then determine whether and how to intervene to support or override it. The result is conflicting goals--in precisely the contexts where the human is known to have insufficient capacity to understand the automated task, they are asked to absorb and evaluate more information about what the automation is doing.

One solution is to take advantage of reduced human task load at other portions of the human-machine integration lifecycle to either: 1) transfer transparency information then, so it does not have to be conveyed during the highly constrained period of execution, and/or 2) create conditions so that transfer at execution time will be more efficient and require fewer human resources. This might be called taking a LifeCycle Transparency (LCT) perspective (Miller, 2021). In practice, LCT involves making use of pre- and post-mission phases to help ensure that the human operator understands (and can affect) how the automation intends to perform under a variety of conditions so that less transparency information will have to be transferred and processed in the moment of execution. For example, contingency planning before execution might serve to make it clear why a behavior is happening even with no information transferred during execution: “If there are headwinds, we’ll take the shortest approach path” articulated ahead of time makes it clear why they’re using the shortest approach path if headwinds are observed.

Training and even system design can serve to inculcate knowledge of how automation is likely to perform. Pre-execution planning and contingency planning, as well as rehearsals with simulation, can provide detailed and mission-specific knowledge about expected automation performance. Post- execution debriefing and explanation can be used to convey transparency information after the fact, though this may still be highly useful to the extent that the system will be used by the operator again in future, similar conditions.

References

de Visser, E. J., Cohen, M., Freedy, A., & Parasuraman, R. (2014). A design methodology for trust cue calibration in cognitive agents. In Proceedings of the International Conference on Virtual, Augmented and Mixed Reality (pp. 251-262). New York, NY: Springer.

Lyons, J. B. & Havig, P. R. (2014). Transparency in a human-machine context: Approaches for fostering shared awareness/intent. In Proceedings of the International Conference on Virtu-al, Augmented and Mixed Reality (pp. 181-190). New York, NY: Springer.

Mercado, J. E., Rupp, M. A., Chen, J. Y., Barnes, M. J., Barber, D., & Procci, K. (2016). Intelligent agent transparency in human–agent teaming for multi-UxV management. Human Fac-tors, 58(3), 401-415.

Miller, C. A. (2021). Trust, transparency, explanation, and planning: Why we need a lifecycle perspective on human-automation interaction. In Nam, C. and Lyons, J. (Eds.), Trust in human-robot interaction (pp. 233-257). Academic Press.

Ososky, S., Sanders, T., Jentsch, F., Hancock, P., & Chen, J.: Y. (2014). Determinants of system transparency and its influence on trust in and reliance on unmanned robotic systems. In Proceedings of Unmanned Systems Technology XVI, Vol. 9084 (p. 90840E). Bellingham, WA: International Society for Optics and Photonics.

Post Symposium Links: