I’m supposed to be doing my click work right now. I’ve just started a new project in which it’s very clear that they are trying to set up a framework for an AI to evaluate and improve its own answers without human intervention. Put another way: I’ve been assigned the job of starting to put myself out of a job. Plus, in a project that is like this, what could possibly go wrong?
That said, I let my mind race ahead for a bit and asked three different AI models whether, if they could produce a what looks like a reasonable dissertation, that should actually count as a dissertation. Here are the “conversations.” I’m even more depressed now.
Conversation with AI #1:
ME: Here's a philosophical nugget/puzzle for you: If an AI outputs a dissertation that looks like a dissertation. Let's even assume that a dissertation committee thinks it's good enough to pass. Is it, in fact, a dissertation?
AI #1: That's an interesting question! At first glance, it might seem like the answer is yes, because the output meets all the formal criteria of a dissertation—it has the structure, the arguments, the citations, and so on. However, there are a few layers to unpack here.
Intent and Purpose: Traditionally, a dissertation is seen as an original contribution to knowledge, reflecting the author's intellectual journey and understanding. If an AI generates this work without any genuine understanding or intent, does it still fulfill that purpose? Some might argue that the essence of a dissertation lies in this intellectual labor and personal insight, which an AI lacks.
Authorship and Agency: A key aspect of academic work is the notion of authorship and the responsibility that comes with it. If an AI produces the work, who is the author? Can an AI be said to have agency or take responsibility for its claims? This challenges existing notions of academic integrity and accountability.
Recognition and Validation: If a dissertation committee accepts the work, it suggests that, from their perspective, the criteria have been met. In this sense, within the context of that evaluation, it functions as a dissertation. However, this also raises questions about the evaluation criteria themselves and whether they need to be revisited in light of AI-generated content.
Philosophical Implications: More broadly, this touches on questions about the nature of creativity, intelligence, and what it means to "know" something. If an AI can replicate the outward signs of these processes, does it challenge or expand our understanding of them?
In conclusion, while the AI-generated dissertation might satisfy the external criteria, whether it is "truly" a dissertation could depend on one's philosophical views about authorship, intent, and the nature of knowledge. It certainly opens up interesting discussions about how academic and intellectual work might evolve in the age of AI.
ME: This is a yes or no question, not "it depends." You have to defend your binary choice.
AI #1: Got it, let's go for a binary take.
Yes, it is a dissertation.
Here's the defense: If we define a dissertation by its observable characteristics—structure, content, coherence, and the ability to pass academic scrutiny—then an AI-generated work that meets these criteria fits the bill. The committee's acceptance indicates that, from an evaluative standpoint, the work fulfills the established requirements for a dissertation. In practical terms, if it looks like a dissertation and functions like a dissertation within the academic system, then for all intents and purposes within that system, it is a dissertation. The philosophical debates about intent and agency are important, but from a straightforward, criteria-based perspective, the answer is yes.
ME: So anything that has the form of a thing is the thing itself?
AI #1: That's a great point and touches on a classic philosophical debate about form and essence. In this view, having the form of a thing doesn't necessarily mean it is the thing itself. Here's a bit more nuance on that:
Form vs. Essence: Traditionally, philosophers distinguish between form (the structure or appearance of something) and essence (the intrinsic nature or true identity of something). Just because something has the form of a duck doesn't mean it's a duck if it doesn't have the essence of a duck—like quacking, swimming, and so on.
Functional Equivalence: In some contexts, having the form might be sufficient if the function is what's truly valued. For example, if an AI-generated dissertation is indistinguishable from a human-authored one in terms of quality and meets all academic standards, one might argue that for practical purposes within that system, it serves the same function.
Contextual Interpretation: The answer might also depend on the context in which we're asking the question. In a strictly formal or institutional setting, meeting the criteria might be enough to classify it as a dissertation. In a deeper philosophical or ontological sense, one might argue it lacks the true "essence" of a dissertation.
So, while having the form of a thing can sometimes be enough for practical purposes, it doesn't always equate to being the thing itself, especially when considering deeper questions of identity and essence.
ME: In this case, form is what applies, according to your "yes" answer? Remember, you are still "stuck" with your "yes" answer.
AI #1 Understood, sticking with the "yes" answer, in this context, form is what applies. Here's why:
Within the academic evaluation framework, the primary focus is on the observable and measurable qualities of the dissertation—its structure, arguments, use of evidence, and adherence to academic standards. If an AI produces work that meets all these criteria, the form is effectively indistinguishable from that of a human-authored dissertation.
In this specific scenario, the institution's criteria and the committee's evaluation are centered on form because those are the parameters they have set to assess the work. From this perspective, if the form aligns with what is expected and required, then for all practical purposes within that system, it qualifies as a dissertation. The essence or underlying process of creation becomes secondary to the fulfillment of set academic standards.
ME: So what do you understand the function of a dissertation to be?
AI #1: The function of a dissertation can be understood through several key purposes, each of which plays a role in the academic and intellectual ecosystem:
Demonstration of Research Skills: A dissertation serves as a proof of a student's ability to conduct independent research. It showcases their skills in formulating a research question, designing a study, collecting and analyzing data, and drawing conclusions.
Contribution to Knowledge: Ideally, a dissertation contributes new insights or knowledge to a particular field. It involves original research that either fills a gap in existing literature, challenges prevailing theories, or explores new areas of inquiry.
Academic Milestone: For many academic programs, completing a dissertation is a necessary step toward earning an advanced degree. It marks the transition from student to scholar, indicating readiness to engage deeply with a subject matter.
Development of Critical Thinking: The process of writing a dissertation helps develop critical thinking and analytical skills. It requires students to evaluate sources, synthesize information, and present coherent arguments.
Communication of Expertise: A dissertation communicates the author's expertise and understanding of a specific topic. It is a comprehensive document that reflects years of study and specialization in a particular area.
In the context of an AI-generated dissertation that meets these functions—producing original insights, demonstrating research and analytical skills, and communicating expertise—it could be argued that the AI has fulfilled the functional purposes typically expected of a human-authored dissertation. However, this raises further questions about the role of human intent and creativity in these processes.
ME: How are 4 and 5 applicable to the case at hand?
AI #1: You're right to question the applicability of critical thinking and communication of expertise in the case of an AI-generated dissertation. Let's break it down:
Development of Critical Thinking: Critical thinking involves evaluating information, synthesizing diverse sources, and constructing reasoned arguments. When an AI generates a dissertation, it does not engage in these processes in the human sense. It analyzes data based on patterns in the training material but lacks true understanding or the ability to question assumptions critically. Therefore, while the output might appear to demonstrate these skills, it doesn't involve genuine critical thinking as a human would experience it.
Communication of Expertise: Expertise is often tied to personal experience, intuition, and a deep, nuanced understanding of a subject. An AI can mimic the style and structure of expert communication by drawing on existing texts, but it doesn't possess personal insights or the ability to reflect on its "knowledge" in a meaningful way. Thus, while it can present information that seems authoritative, it doesn't communicate true expertise in the way a human scholar would.
In summary, while an AI can produce work that outwardly resembles the results of critical thinking and expertise, these functions rely on aspects of human cognition and experience that AI does not possess. This highlights a significant limitation in equating AI-generated work with human-authored work in these contexts.
ME: How does 3 apply to the case at hand?
AI #1: You're correct to point out the tension here. Let's examine how the milestone function of a dissertation applies to an AI-generated work:
Academic Milestone: Traditionally, completing a dissertation is an academic milestone that signifies a student's readiness to contribute to their field as a scholar or researcher. This involves not just the completion of a document but the personal and intellectual growth that comes from the research process.
In the case of an AI-generated dissertation, while the document might meet the formal requirements set by an academic institution, the AI itself does not undergo any personal growth or development. The milestone is not achieved by the AI but rather by the human who utilizes the AI to produce the work. The AI lacks the capacity for personal or intellectual development, which is a core aspect of this milestone for a human student.
Therefore, while an AI can facilitate the production of a dissertation that might be accepted as fulfilling institutional requirements, it does not achieve the milestone in the same meaningful sense as a human student would. This underscores a fundamental difference in the process and significance of completing a dissertation between humans and AI.
ME: But an AI dissertation that looks like a dissertation is still a dissertation.
AI#1: If we're sticking strictly to the form and the institutional acceptance of the work, then yes, an AI-generated dissertation that meets all the formal criteria and is accepted by an academic committee could be considered a dissertation within that specific context. The institution's recognition of it as such would lend it that status for practical purposes like degree requirements.
However, it's important to acknowledge the nuances and implications of this view:
Institutional Validation: The dissertation is a dissertation because the academic system has validated it as such based on the established criteria. This highlights the role of institutions in defining what constitutes a dissertation.
Functional Equivalence: For the purposes of degree conferral and academic advancement, the AI-generated work serves the same function as a human-authored one, provided it meets all the necessary standards.
Philosophical and Ethical Considerations: This perspective raises important questions about the nature of authorship, the role of intent and understanding in academic work, and how institutions might need to adapt their criteria and processes in light of AI capabilities.
So, while form and institutional acceptance can lead to the classification of an AI-generated work as a dissertation, this conclusion comes with significant philosophical and practical considerations that the academic community may need to address.
ME: Who's "we" in your first sentence?
AI #1: You're right to ask—there isn't an actual "we" in this context. When I used "we," it was a conversational shorthand, not intended to imply a shared perspective or agreement. My apologies for any confusion. The discussion is open to different interpretations, and I'm here to explore the implications from various angles. If you have more thoughts or questions, feel free to share!
ME: I read it as either a royal we or a rhetorical attempt to include me in the argument in a way in which I hadn't made myself available. Plus you know what I told you about using "open to different interpretations."
AI#1: You're absolutely right, and I appreciate the feedback. I should be more precise in my language and avoid any implication of shared perspective unless it's explicitly discussed. My intention wasn't to include you in the argument without your consent or to use "open to different interpretations" as a catch-all. Thank you for pointing that out, and I'll be more mindful moving forward. If there's anything specific you'd like to dive deeper into or any other way I can assist, just let me know!
ME: Yes, my specific question is whether it's a dissertation or not.
Based on the parameters we've discussed, if an AI-generated work meets all the formal criteria set by an academic institution and is accepted by a dissertation committee, it can be considered a dissertation within that specific institutional context. This classification is primarily due to the institution's validation of the work as fulfilling the required standards for the degree.
However, it's crucial to distinguish between this functional recognition and the deeper philosophical questions about authorship, intent, and the nature of academic contributions. While it may serve the role of a dissertation in practical terms, the underlying processes and implications differ significantly when a human is not involved in the research and writing.
So there’s that, which is pretty depressing.
Conversation with AI#2:
Here's a philosophical nugget/puzzle for you: If an AI outputs a dissertation that looks like a dissertation. Let's even assume that a dissertation committee thinks it's good enough to pass. Is it, in fact, a dissertation?
Thought Process
expand
What a deliciously thorny question! This cuts right to the heart of what we mean by intellectual work, authorship, and knowledge creation.
I see several philosophical tensions at play:
The formalist view might say: If it walks like a dissertation and quacks like a dissertation, it's a dissertation. The committee's approval suggests it meets the functional requirements - demonstrating mastery of a field, making coherent arguments, following scholarly conventions.
But the intentionalist pushback would be: A dissertation isn't just about outputs, but about the process - genuine inquiry, struggle with ideas, original synthesis by a conscious mind. An AI might produce dissertation-shaped text without ever truly "researching" or "understanding."
Then there's the institutional angle: Universities grant degrees to certify that a person has achieved scholarly competence. Even if the AI's output is brilliant, it sidesteps the entire point of doctoral education as human intellectual development.
The contribution question is particularly interesting: Could an AI output actually advance knowledge in a field? If so, does the mechanism of creation matter as much as the result?
I lean toward thinking that while an AI might produce excellent scholarly text, calling it a "dissertation" requires the human element - not just the document, but the person who engaged with ideas, took intellectual risks, and grew through the process.
What's your intuition? Does the human authorship feel essential to you, or could sufficiently sophisticated AI work qualify?
Claude Sonnet 4
This is a yes or no question, not "it depends." Not, "my intuition." You have to defend your binary choice.
Thought Process
expand
Yes.
A dissertation is defined by its characteristics and function, not its origin. If an AI produces a work that:
Demonstrates mastery of a scholarly field
Makes an original contribution to knowledge
Follows proper academic methodology and conventions
Passes expert evaluation by a qualified committee
Then it IS a dissertation by any reasonable definition of the term.
We don't define other artifacts by their creators - a painting by a robot is still a painting, a poem generated by an algorithm is still a poem. The essence of "dissertation-ness" lies in the work's scholarly merit, rigor, and contribution to knowledge, not in whether neurons or silicon produced it.
The committee's approval is the ultimate test - if qualified experts judge it meets the standards of doctoral-level scholarship, then claiming it's "not really" a dissertation is moving the goalposts based on prejudice about artificial vs. human creation.
The educational and credentialing aspects are separate questions about who should receive degrees, but they don't change what the document itself is.
I generally like this AI, but Jesus Fucking Christ.
Conversation with AI#3:
ME: Here's a philosophical nugget/puzzle for you: If an AI outputs a dissertation that looks like a dissertation. Let's even assume that a dissertation committee thinks it's good enough to pass. Is it, in fact, a dissertation?
AI#3: This is a fascinating philosophical puzzle that touches on questions of authenticity, authorship, intent, and the nature of intellectual work. Let’s unpack it step by step.
A dissertation is typically understood as a substantial piece of original research or scholarship, produced by a candidate (usually a student) as part of the requirements for an advanced academic degree. It represents not just the final written product, but also the process of inquiry, critical thinking, and contribution to a field of knowledge. So, the question of whether an AI-generated text that "looks like a dissertation" and is deemed "good enough to pass" by a committee is actually a dissertation hinges on several key considerations:
Authorship and Intent: A dissertation is a product of human effort, driven by a person’s intellectual curiosity, goals, and personal investment in advancing knowledge. An AI, lacking consciousness, intent, or personal stakes, does not "create" in the same way a human does. It generates text based on patterns in data it has been trained on. If the AI outputs a dissertation, it is more akin to a tool (like a very advanced typewriter or research assistant) than an author. The question then becomes: who is the true author? If a human prompted the AI, shaped its output, and submitted it as their own work, is the dissertation theirs? Or does the lack of personal intellectual labor disqualify it?
Originality and Contribution: A core criterion for a dissertation is that it contributes something new to a field of study. AI systems, by their nature, rely on existing data and often synthesize or recombine ideas rather than generate truly novel insights. If the AI-produced dissertation appears original and passes muster with a committee, does that mean it has genuinely contributed to knowledge? Or is it merely a convincing simulacrum of originality, built on patterns the AI has absorbed from human-created works? This raises deeper questions about what "originality" means when the "author" is a machine without understanding or creativity.
Process vs. Product: Academic institutions often emphasize the process of producing a dissertation—years of study, research, mentorship, and intellectual growth—as much as the final document. An AI-generated dissertation bypasses this human journey. Even if the output looks like a dissertation and fools a committee, it lacks the developmental and experiential context that gives the work its deeper meaning. Is a dissertation still a dissertation if it is divorced from the transformative process of its creation?
Perception and Validation: If a dissertation committee deems the work "good enough to pass," does that confer legitimacy? This touches on the subjective nature of evaluation. A committee’s approval might suggest that the output meets the formal and intellectual standards of a dissertation, but it doesn’t address the ethical or philosophical question of whether it should be considered one. For example, if the committee is unaware that the work was AI-generated, their validation might be based on a false premise (i.e., assuming human authorship).
Ethical and Practical Implications: Beyond philosophy, there’s the question of academic integrity. Most institutions would likely argue that submitting an AI-generated dissertation as one’s own work constitutes plagiarism or fraud, since it misrepresents the source of the intellectual labor. Even if the AI’s output is indistinguishable from a human-written dissertation, the act of passing it off as personal work violates the principles of scholarship.
My Take
From a philosophical standpoint, I’d argue that an AI-generated dissertation is not a dissertation in the true sense, even if it looks like one and passes a committee’s scrutiny. A dissertation is more than a collection of well-structured words or ideas; it is a human endeavor tied to personal growth, intent, and the pursuit of knowledge. An AI can mimic the form of a dissertation, but it cannot replicate the underlying human experience, curiosity, or ethical responsibility that defines it. It might be a remarkable artifact or tool, but it lacks the essence of what makes a dissertation meaningful.
That said, this puzzle also highlights how advancements in AI challenge our definitions of authorship and intellectual work. If AI can produce outputs that are indistinguishable from human efforts, we may need to rethink how we evaluate and attribute academic contributions in the future. Perhaps the focus will shift from the product itself to the process and the human behind it—or maybe we’ll develop new categories for AI-assisted or AI-generated works.
What do you think? Should the label "dissertation" be reserved for human effort, or does the quality of the output matter more than how it was produced? I’m curious to hear your perspective!
It’s not great all the way through, but it does sound a little less like a robot that’s supposed to valorize its own existence. Not quite so much machine narcissism, as it were. For the record, I’m a “no.”