Presentation
Navigating Uncharted Waters: The AI Revolution in Human Factors Workflows
DescriptionArtificial Intelligence (AI), and specifically Large Language Models (LLMs), are quickly proving to be the most disruptive technology of the modern era. The transformative power of such LLMs promises to drive multiple levels of societal change as it is being applied widely across sectors with numerous spill-over effects. AI is on the precipice of a massive wave, given the technological, sociological, and economic factors that position this novel technology at the forefront of change. The cost of running advanced AI models decreased by over 280-fold between November 2022 and October 2024, making AI more accessible and efficient for organizations and individuals to utilize. The majority of undergraduate and graduate students worldwide (86%) use AI tools for their studies. Hence, it is foreseeable that most professionals will rely on LLMs in some capacity in the near future, given the promises of efficiency and streamlined production.
The enthusiasm surrounding this area is well-founded, and the potential benefits are substantive. A recent survey of 26 nations shows that global optimism about AI has increased as the share of people who view AI products and services as more beneficial than harmful rose to 55 % in 2024. Additionally, the number of AI-related publications have more than doubled between 2013 and 2023, rising from about 102,000 to over 242,000, evidencing the growing research and academic interest. Organizational adoption is also on the rise, with over 73% of organizations worldwide currently using or piloting AI in their core functions. Finally, even regulatory bodies, which have historically been slow to adopt new technologies, have increasingly accepted them, as evidenced by the almost exponential increase in approved AI-enabled medical devices by the FDA between 2015 and 2025, which nearly quintupled from 221 to over 1200 according to the agency's database.
However, without guardrails, various detrimental effects can be caused by technologies like AI and LLMS. For example, technological overreliance can lead to patient deaths, as happened with the Therac-25, a computer-controlled radiation therapy machine which, between 1985 and 1987, delivered massive radiation overdoses (up to 100x prescribed) to patients due to software errors and operators' trust in the computer readouts, which masked malfunctions. Lack of oversight on algorithm-driven systems, such as Automated Insulin Delivery (AID) systems, may harm patients, causing hypoglycemia or hyperglycemia, hospitalizations, or even death, due to automatic Insulin dosing adjustment based on malfunctioning Continuous Glucose Monitoring (CGM) sensors. These examples illustrate some of the unforeseen consequences and drawbacks that may accompany novel technological advancements, which brings us to the proposed discussion panel.
We want to call upon the HFES community to anticipate and address potential threats associated with AI, as it redefines roles across every industry, including our own as human-centered professionals. As humanity grapples with new AI-human teams in every sector, human factors engineers (HFEs) are particularly well-positioned to characterize, analyze, and understand human-AI systems for the fulfillment of daily tasks.
This discussion panel aims to facilitate a thoughtful discussion on the strengths, limitations, and functions of HFE-LLM interactions as we integrate them into our daily activities and research strategies, enabling human factors engineers to be deliberate in our interactions, anticipate potential negative consequences, and protect our field accordingly.
The proposed discussion panel, expected to last approximately 60-90 minutes, will follow a typical discussion format, with moderators asking panelists open-ended questions about their experiences integrating AIs, specifically LLMs, into their workflows. At the beginning of the panel, the session will also include an audience participation portion through the distribution of a very brief online survey via a real-time polling app, regarding their use of AI and LLMs (How often do you use LLMs for work? What type of tasks do you usually use it for?, etc).
The moderators will also introduce the Complementary Ability-Function-Role (CAFR) framework proposed by Yehong Zhou and Chun-Hsien Chen (2025) as a launchpad for discussing HFE-LLM collaboration. By utilizing this framework, we intend to exhibit, challenge, and discuss the core competencies of LLMs and the core roles of HFEs as they collaborate on divergent and convergent thinking throughout distinct creative phases: Discover, Define, Develop, Deliver. Throughout the discussion, one of the moderators will categorize the audience survey responses and panelist responses into these phases, which will be revealed towards the end of the session and used to prompt further discussion and audience questions.
The following five questions will serve as the basis for the discussion. Additional prompting questions are included to help explore further, as needed, along with a few bonus questions, which may be asked as time allows.
INTRODUCTIONS: What is your background in 1-2 sentences? What does your relationship with AI and your professional workflow look like?
1. Where is the most success you’ve had with AI?
What is your favorite application of AI? Why?
What value did AI bring to this application? E.g., time savings, re-framing, iciting new ideas .
2. When has AI functionally “Gone wrong” for you, or how did it fail to add value?
Where do you not trust AI?
In what ways has AI failed?
What are the pain points when integrating AI into your workflow?
3. What challenges do you see organizations face adopting AI?
In what ways do you foresee AI supporting your future role as an HFE?
What privacy (e.g., PHI, client data, etc.) considerations or precautions must be taken?
How does the lack of reliability or accuracy due to AI hallucinations impact adoption?
How do you envision the knowledge of entry-level vs senior-level knowledge impacting their use of AI and the collaboration outputs?
Can entry-level professionals use AI with the same proficiency as mid & late career professionals?
Is there a perceived role obsolescence? What would role obsolescence look like?
4. How can organizations adopt AI responsibly?
What are your biggest concerns?
How can organizations incentivize technological equity across generational & socio-economic users?
What are the environmental concerns?
5. AI is often framed as an advantage; are there ways you see it as a disadvantage?
In what ways would non-AI-assisted humans thrive or struggle?
What would a world look like where a subset of professionals use AI while others remain?
What do you see as “watch-outs” to be aware of? Where are you wary/concerned?
WRAP UP & RECAP
At the end of the session, panelists will be invited to provide closing remarks. The discussed applications of AI will be presented on a display, contextualized for the CAFR framework, highlighting the core human and AI competencies that were exemplified. Finally, results from the audience live survey will be projected to map the overview of how attendees are utilizing AI in their workflows.
Audiences attending the discussion panel will leave with the following key takeaways:
Learn successful ways in which professionals are currently using LLMs and become aware of the common pitfalls.
Consider the CAFR framework as a way to re-imagine their next AI-human collaboration endeavour, by targeting the function and each agent’s strengths, taking full advantage of both AI and humans’ core competencies to maximize the best team interaction.
Embrace intentional usage of LLMs as tools rather than a replacement for human factors engineers’ work.
Panel schedule:
10-minute - panel topic, panelists' introduction
5 minute - Introduction of CAFR framework
30 - 50-minute - Panelist questions
5 - 10-minute - Wrap up and recap, putting the discussed ideas in CAFR framework
10-minute - Questions from the audience
The enthusiasm surrounding this area is well-founded, and the potential benefits are substantive. A recent survey of 26 nations shows that global optimism about AI has increased as the share of people who view AI products and services as more beneficial than harmful rose to 55 % in 2024. Additionally, the number of AI-related publications have more than doubled between 2013 and 2023, rising from about 102,000 to over 242,000, evidencing the growing research and academic interest. Organizational adoption is also on the rise, with over 73% of organizations worldwide currently using or piloting AI in their core functions. Finally, even regulatory bodies, which have historically been slow to adopt new technologies, have increasingly accepted them, as evidenced by the almost exponential increase in approved AI-enabled medical devices by the FDA between 2015 and 2025, which nearly quintupled from 221 to over 1200 according to the agency's database.
However, without guardrails, various detrimental effects can be caused by technologies like AI and LLMS. For example, technological overreliance can lead to patient deaths, as happened with the Therac-25, a computer-controlled radiation therapy machine which, between 1985 and 1987, delivered massive radiation overdoses (up to 100x prescribed) to patients due to software errors and operators' trust in the computer readouts, which masked malfunctions. Lack of oversight on algorithm-driven systems, such as Automated Insulin Delivery (AID) systems, may harm patients, causing hypoglycemia or hyperglycemia, hospitalizations, or even death, due to automatic Insulin dosing adjustment based on malfunctioning Continuous Glucose Monitoring (CGM) sensors. These examples illustrate some of the unforeseen consequences and drawbacks that may accompany novel technological advancements, which brings us to the proposed discussion panel.
We want to call upon the HFES community to anticipate and address potential threats associated with AI, as it redefines roles across every industry, including our own as human-centered professionals. As humanity grapples with new AI-human teams in every sector, human factors engineers (HFEs) are particularly well-positioned to characterize, analyze, and understand human-AI systems for the fulfillment of daily tasks.
This discussion panel aims to facilitate a thoughtful discussion on the strengths, limitations, and functions of HFE-LLM interactions as we integrate them into our daily activities and research strategies, enabling human factors engineers to be deliberate in our interactions, anticipate potential negative consequences, and protect our field accordingly.
The proposed discussion panel, expected to last approximately 60-90 minutes, will follow a typical discussion format, with moderators asking panelists open-ended questions about their experiences integrating AIs, specifically LLMs, into their workflows. At the beginning of the panel, the session will also include an audience participation portion through the distribution of a very brief online survey via a real-time polling app, regarding their use of AI and LLMs (How often do you use LLMs for work? What type of tasks do you usually use it for?, etc).
The moderators will also introduce the Complementary Ability-Function-Role (CAFR) framework proposed by Yehong Zhou and Chun-Hsien Chen (2025) as a launchpad for discussing HFE-LLM collaboration. By utilizing this framework, we intend to exhibit, challenge, and discuss the core competencies of LLMs and the core roles of HFEs as they collaborate on divergent and convergent thinking throughout distinct creative phases: Discover, Define, Develop, Deliver. Throughout the discussion, one of the moderators will categorize the audience survey responses and panelist responses into these phases, which will be revealed towards the end of the session and used to prompt further discussion and audience questions.
The following five questions will serve as the basis for the discussion. Additional prompting questions are included to help explore further, as needed, along with a few bonus questions, which may be asked as time allows.
INTRODUCTIONS: What is your background in 1-2 sentences? What does your relationship with AI and your professional workflow look like?
1. Where is the most success you’ve had with AI?
What is your favorite application of AI? Why?
What value did AI bring to this application? E.g., time savings, re-framing, iciting new ideas .
2. When has AI functionally “Gone wrong” for you, or how did it fail to add value?
Where do you not trust AI?
In what ways has AI failed?
What are the pain points when integrating AI into your workflow?
3. What challenges do you see organizations face adopting AI?
In what ways do you foresee AI supporting your future role as an HFE?
What privacy (e.g., PHI, client data, etc.) considerations or precautions must be taken?
How does the lack of reliability or accuracy due to AI hallucinations impact adoption?
How do you envision the knowledge of entry-level vs senior-level knowledge impacting their use of AI and the collaboration outputs?
Can entry-level professionals use AI with the same proficiency as mid & late career professionals?
Is there a perceived role obsolescence? What would role obsolescence look like?
4. How can organizations adopt AI responsibly?
What are your biggest concerns?
How can organizations incentivize technological equity across generational & socio-economic users?
What are the environmental concerns?
5. AI is often framed as an advantage; are there ways you see it as a disadvantage?
In what ways would non-AI-assisted humans thrive or struggle?
What would a world look like where a subset of professionals use AI while others remain?
What do you see as “watch-outs” to be aware of? Where are you wary/concerned?
WRAP UP & RECAP
At the end of the session, panelists will be invited to provide closing remarks. The discussed applications of AI will be presented on a display, contextualized for the CAFR framework, highlighting the core human and AI competencies that were exemplified. Finally, results from the audience live survey will be projected to map the overview of how attendees are utilizing AI in their workflows.
Audiences attending the discussion panel will leave with the following key takeaways:
Learn successful ways in which professionals are currently using LLMs and become aware of the common pitfalls.
Consider the CAFR framework as a way to re-imagine their next AI-human collaboration endeavour, by targeting the function and each agent’s strengths, taking full advantage of both AI and humans’ core competencies to maximize the best team interaction.
Embrace intentional usage of LLMs as tools rather than a replacement for human factors engineers’ work.
Panel schedule:
10-minute - panel topic, panelists' introduction
5 minute - Introduction of CAFR framework
30 - 50-minute - Panelist questions
5 - 10-minute - Wrap up and recap, putting the discussed ideas in CAFR framework
10-minute - Questions from the audience
Moderators
Sr. Human Factors Engineer
Panelists
Sr. Human Factors Engineer
Event Type
Discussion Panel
TimeMonday, March 233:30pm - 4:30pm EDT
LocationGramercy
Medical and Drug Delivery Devices