Performance and Competence

There’s a good interview with Rodney Brooks about artificial intelligence at IEEE Spectrum. I was struck especially by an idea that he has been promoting for some time, namely, that the way we generalize from performance to competence in humans cannot be applied to machines. If a human being is able to perform some cognitively sophisticated task we are warranted to assume that they can perform many other sophisticated tasks, whereas when a machine does something similar, we can’t use that as an indication of some underlying general competence. The reasons for this are nicely explained, as Brooks points out, by a recent blogpost by Stephen Wolfram, which everyone who wants to have an opinion about ChatGPT should probably read. The point is that once you know how ChatGPT does what it does, you can’t attribute sentience to it just because it uses language in an almost human way. By contrast, when a human body uses language, it is a clear indication of a deep and broad array of cognitive functions. As far as we know, human beings can’t use language without being sentient (and conscious and intelligent). And once we know this about them, we know a great deal about what they’re capable of, just from watching and listening to them speak.

This is why the sudden emergence of language models has been such a shock to many of us in academia. We have been accustomed to taking the performance of students in their writing assignments as a good indication of their knowledge competence. Consider the following paragraph, which I easily coaxed out of ChatGPT (3.5) in a few minutes:

According to Weick (1995), sensemaking is a reflective and retrospective process. Weick (1996, p. 54) defines sensemaking as an "ongoing retrospective process of forming images that justify what people are doing." It is a cognitive mechanism that individuals and organizations employ to comprehend complex and ambiguous situations. Sensemaking involves looking back at events and experiences, constructing meanings, and forging understandings. By interpreting and attaching significance to events based on existing mental frameworks, sensemaking allows for the creation of order and coherence amidst uncertainty. Weick emphasizes that sensemaking is an ongoing and iterative process, continuously evolving as new information and perspectives emerge. It encompasses the act of storytelling and constructing narratives to give meaning to our experiences. Sensemaking extends beyond the individual, serving as a social process that takes shape through interactions and conversations. It serves as a means to reduce ambiguity, facilitating decision-making in intricate and turbulent environments. Weick highlights the importance of sensemaking in generating and maintaining a coherent understanding of the world around us. Additionally, sensemaking plays a pivotal role in sensegiving, the process of effectively communicating and influencing others' understanding of a given situation. By elucidating Weick's perspective and incorporating his quote, it becomes evident that sensemaking is a fundamental tool for navigating complexity and making informed choices.

I gave it the key sentence “Sensemaking is a retrospective process”, the source (Weick 1995), and the quoted definition (from memory, wrong year, wrong page). I had it think about how to pitch the key sentence, brainstorm some sentences for the body of the paragraph, and then compose one coherent paragraph that helped the reader understand Weick. After a little more back and forth, the result was the above paragraph. At no point did I edit the paragraph. That is, none of the sentences are mine.

I then asked it to “construct a one-paragraph description (a story) of sensemaking in action to illustrate these ideas”. It gave me this:

Sarah, a project manager, found herself facing a daunting challenge. The unexpected departure of a key team member had left the project in disarray, with looming deadlines and a complex web of interdependent tasks. Engaging in sensemaking, Sarah began her reflective journey. She revisited past team meetings, scrutinized project documentation, and engaged in conversations with remaining team members. As she immersed herself in this retrospective process, images started forming in her mind, providing justifications for the team's current situation and possible paths forward. Sarah realized that the departure of the team member had disrupted the workflow, leading to miscommunication and overlapping responsibilities. Drawing upon her mental frameworks, she began constructing a narrative, identifying the root causes and envisioning potential solutions. She engaged the remaining team members in discussions, seeking their perspectives and insights. Together, they collaboratively made sense of the situation, aligning their understandings and identifying actionable steps. Through this ongoing and iterative sensemaking process, Sarah and her team regained clarity, rebuilt cohesion, and developed a shared understanding of the project's direction. Sensemaking not only empowered Sarah to navigate complexity but also fostered a collective sense of purpose, enabling the team to move forward with confidence and resolve.

Now, there is perhaps nothing Earth-shatteringly brilliant about these two paragraphs. But, in the old days, working on the assumption that the student had actually written them, I would have inferred a relatively deep underlying competence in them. The student is able to present an understanding of what they have read; indeed, they could presumably not have written that paragraph without reading Weick. Also, they are able to imagine or observe a reality that corresponds to the concepts they have introduced. It could be argued that the “imaginative” or “empirical” material is a bit abstract. But it didn’t take me much more coaxing to get ChatGPT to give something more concrete:

But the piles of data on Sarah's desk weren't sufficient. She craved the human insights that could illuminate the path forward. With determination, she picked up the phone and called the remaining team members, inviting them for a face-to-face discussion.

In a small meeting room, Sarah sat across from John, a seasoned team member who had been with the project since its inception. She leaned forward, her voice earnest, "John, I need your help. I want to understand what went wrong and how we can fix it. Tell me about your experiences and frustrations."

John's brow furrowed as he recalled the recent challenges. "Well, Sarah, since Mark left, communication has become a nightmare. We used to have regular team huddles where we shared updates, but without him, things have fallen apart. Everyone seems unsure of their responsibilities, and tasks are slipping through the cracks."

Sarah nodded, her notepad ready to capture every valuable insight. "I see. Can you provide specific instances where miscommunication occurred? Any ideas on how we can improve the clarity of roles?"

John scratched his head thoughtfully. "One example is the client meeting last week. We were all working on different aspects of the project, but nobody knew what the others were doing. During the meeting, the client asked questions that none of us could answer. It was embarrassing and made us look unprepared."

Sarah scribbled notes, her mind racing to identify patterns. "So, improved communication and clearer role delineation are crucial. What suggestions do you have for addressing these issues, John?"

John leaned back, his expression thoughtful. "We need a central hub for information, where everyone can access updates, share progress, and see the big picture. Regular team meetings should be reinstated, so we can align our efforts and address any roadblocks together. And we need to clarify each team member's responsibilities and establish clear lines of communication."

Sarah's eyes lit up with excitement. "Those are excellent suggestions, John. I truly appreciate your insights. Let's discuss these ideas with the rest of the team and see how we can implement them effectively."

As Sarah continued her conversations with other team members, she encountered similar stories of miscommunication, frustration, and potential solutions. Jennifer, a detail-oriented analyst, shared her experiences with overlapping responsibilities that resulted in redundant work. David, a junior developer, expressed his struggles to navigate the project without clear guidance.

Armed with these specific insights, Sarah synthesized the collective wisdom and proposed tangible actions during their next team gathering. The air buzzed with anticipation as Sarah laid out the proposed central hub for information sharing, the reinstatement of regular team meetings, and the need for role clarity. Dialogue ensued, with team members offering further suggestions and refining the proposed solutions.

Through these concrete conversations and shared experiences, Sarah and her team co-created a roadmap for moving forward. They set deadlines for implementing the new communication practices, assigned responsibilities for maintaining the central hub, and established a process for clarifying roles and expectations.

The team members left the meeting feeling heard, energized, and part of a collective effort. Concrete steps had been taken, thanks to the specific insights shared during their conversations. The path forward was clearer, and a sense of unity and purpose permeated the team.

Again, this isn’t profound stuff. But a human student who could do this, either through imagination or observation, would be showing they can do many other things. I keep saying “imagination or observation” because, in the case of a human student, such an account may be presented as made up or as based on data. We expect the student to be honest about such things, and if the account really were empirical we would expect them to be able to write a plausible statement of method. From this performance (of writing a methods statement) we would again infer all manner of underlying competence.

But ChatGPT can give us such a statement too. In a perhaps unnecessary attempt to avoid certain “safety” measures, I prompted it to “imagine that the account is the result of participant observation by an ethnographer. Write a paragraph for the methods section in which you imagine how the data the account of Sarah’s sensemaking is based on.” After giving me an explicitly imaginary account of “an ethnographer”, I asked it simply to rewrite it in the first person as a straight statement of method.

As an ethnographer, I immersed myself in the project management context to conduct participant observation and gain insights into Sarah's sensemaking process. Over an extended period, I actively observed team meetings, carefully documenting the dynamics, non-verbal cues, and communication patterns. I took detailed field notes, capturing the atmosphere, interactions, and notable events. Additionally, I engaged in informal conversations with Sarah and other team members, both individually and in group settings, to delve into their experiences, frustrations, and suggestions. These conversations were audio-recorded to ensure accurate representation. To supplement my observations and interviews, I thoroughly reviewed project documentation, including meeting minutes, task assignments, and relevant correspondence. The data collected through participant observation, field notes, interviews, and document analysis were meticulously analyzed and synthesized to construct a comprehensive account of Sarah's sensemaking journey. Through my firsthand involvement and interpretation, I aimed to provide an authentic and nuanced portrayal of the project's context, Sarah's challenges, and the collaborative sensemaking process that unfolded.

I really wish we could still use the writing students do at home to gauge whether they have learned what we were trying to teach them. But examples like this are enough to completely undermine my faith in such performances as indications of actual competence. Going forward, I only want to look at writing from students that was produced off-line. I want to know what they are capable of without the assistance of AI.

* * *

PS. All the outputs in the above post were generated using ChatGPT model 3.5. I gave the long ethnographic passage to 4.0 and asked it to help me write a methods statement. Here’s its first attempt:

Based on the ethnographic description provided, here is a potential methods statement:

This ethnographic study adopts an interpretative, qualitative approach to understanding project management under crisis conditions. The research method includes direct observation and engagement with the research subject, Sarah, a project manager. The research subject was observed in her natural environment at her workplace to understand her behaviors, actions, and interactions in a real-life context.

Sarah's activities were meticulously documented as she navigated through a crisis caused by the sudden departure of a key team member. The observation involved recording her approach to problem-solving, including her reflective analysis of previous records, her communication with team members, her mental processes during decision-making, and her efforts towards group sensemaking and decision-making.

Primary data collection involved the naturalistic observation of Sarah's behaviors, conversations, and interactions, supported by the collection of project-related documents like meeting notes and records. Additionally, informal interviews were conducted with Sarah and her team members to gain further insight into their experiences, frustrations, perspectives, and potential solutions.

The analysis process was inductive and iterative, informed by grounded theory. This process helped to identify patterns, themes, and emergent properties related to team communication, role clarity, and collective sensemaking. These findings were then used to construct a narrative that detailed Sarah's sensemaking process during crisis management.

Sarah's experiences were triangulated with the insights shared by the other team members. This approach allowed for a richer, more nuanced understanding of the crisis situation and how the team navigated through it. The interpretive approach allowed for the examination of the shared sensemaking process, enabling the research to delve deeper into the team's collective resilience and problem-solving approach.

The outcome of this research is intended to provide insights into effective crisis management strategies, with a particular focus on the importance of collaborative sensemaking. It is hoped that these findings will contribute to the broader literature on project management and organizational behavior, as well as provide practical insights for those managing teams under crisis conditions.

I then asked it for a background description of Sarah’s organization.

Sarah's organization, InnoTech, is a leading mid-sized tech company located in San Francisco, California. With a workforce of approximately 500 employees, the company specializes in developing cutting-edge software solutions that address the critical needs of businesses across various sectors, including healthcare, finance, and e-commerce.

InnoTech has a reputation for fostering innovation and technological advancement. It's been in operation for more than a decade and has become well-regarded for its pioneering contributions to the tech industry, primarily through its suite of productivity-enhancing applications and data analytics platforms.

The organization prides itself on its agile work environment and flat hierarchical structure. Teams within InnoTech function as autonomous units, each overseeing different projects with specific goals aligned with the company's broader mission. This structure promotes a high degree of collaboration, enabling cross-functional teams to work seamlessly together.

Sarah's team, known as the Phoenix Project Team, consists of developers, UI/UX designers, quality assurance engineers, and product managers. Their role is to conceptualize, design, develop, and launch new software products aimed at enhancing business productivity.

The company's culture emphasizes continuous learning and improvement, fostering a climate of creativity and risk-taking. It invests significantly in its employees, offering them opportunities for professional growth and development. There's also a strong focus on maintaining a positive work-life balance, reflected in flexible working arrangements and comprehensive employee wellness programs.

However, despite its positive attributes, the organization isn't immune to challenges. As depicted in the ethnographic description, sudden staff departures and project disruptions can put immense pressure on teams, necessitating robust crisis management strategies. It's within this context that Sarah, as a project manager, navigates her role, working to ensure her team effectively overcomes obstacles and delivers on their project objectives.

Leave a Reply

Your email address will not be published. Required fields are marked *