--- title: "AI as Cognitive Prosthetic:\nYou are the art, AI is an interface, the result is value." description: "Moving from answer generation to insight extraction. Agentic systems should augment human thought. They function as high-performance 'cyberpunk chrome' designed to extract and amplify the wisdom already present in your mind." date: "2026-02-04" tags: ["AI & Machine Learning", "Brain-computer interfaces", "Neuroscience", "Agentic AI", "Cognition", "Reasoning & Prompting", "human-AI collaboration", "cognitive augmentation", "tool use"] keywords: "agentic AI, cognitive prosthetic, human-AI collaboration, insight extraction, cognitive augmentation, AI agency, prosthetics, agency preservation" author: "Bartosz Lenart" thumbnail: "/assets/blog/ai-cognitive-prosthetic/ai-cognitive-prosthetic-thumbnail.png" readingTime: "13 min listen" ttsVoiceHeadings: true ---
Get Instant Insight Learn how to treat agentic AI as a cognitive prosthetic - a tool for extracting insights from your own mind rather than outsourcing thought to machines. This framework combines cyberpunk metaphors with cognitive science to help you maintain agency while leveraging AI's computational precision. --- ### The Core Idea - **Prosthetic vs Oracle:** AI should extract insight from *your* mind, not generate answers *for* you - **Chrome Integration:** Like cybernetic augmentation, AI extends human capability while remaining under neural command - **Transformation Through Friction:** The value lies in the struggle of creation, not just the output - **Agency Preservation:** You remain the artist; AI is the precision tool **Framework Purpose:** A mental model for using AI that preserves human agency, maintains cognitive muscle, and treats machines as amplifiers of human wisdom rather than replacements. --- ### The Problem: Outsourcing Agency When we delegate thinking to AI, we risk what Brandon Sanderson identifies as the fundamental loss in AI-generated art: **the transformation that happens through the creative process itself**[^10][^11][^14]. Sanderson argues: "I don't care if the AI can create something that is better than what we can create because it cannot be changed by that creation"[^10]. The act of creation - the struggle, the failure, the refinement - is what transforms the creator. This is what makes us the art. **The Shift:** - **OLD:** AI as Digital Oracle "Give me the answer" - **NEW:** AI as Cognitive Prosthetic "Help me extract my insight" --- ### The Journey Is the Art Sanderson's thesis centers on a profound truth: **art's primary value lies not in the output but in how creating it changes you**[^10][^11][^14][^16]. When you write a novel: - You read hundreds of books that expand your mind - You struggle with structure, developing pattern recognition - You fail repeatedly, building resilience - You refine your voice, discovering who you are The manuscript is the *receipt* of that transformation. AI that generates the output steals the journey[^10][^14]. **This applies beyond traditional art.** When you: - Manually analyze data You develop intuition for patterns - Write code from scratch You understand system architecture - Solve problems iteratively You build mental models The cognitive "friction" is where learning happens. AI that eliminates all friction eliminates growth. --- ### What Makes This Framework Different Unlike treating AI as an autonomous answer-engine, the cognitive prosthetic model changes the relationship: | Answer-Engine (Digital Oracle) | Cognitive Prosthetic (Chrome) | |--------------------------------|-------------------------------| | Provide definitive conclusions | Extract and refine user intuition | | One-shot or linear chat | Recursive, multi-agent loops | | Summarize external data | Synthesize user's seed thoughts | | Success = Output accuracy | Success = Depth of user's insight | | User = Information consumer | User = Process director | **The prosthetic model preserves what Sanderson identifies as essential: you remain the creator, changed by the act of creation.**
--- ## Introduction **Agentic AI is shifting:** away from the "Digital Oracle" that provides answers, toward the "Cognitive Prosthetic," high-performance "cyberpunk chrome" integrated into the human intellectual workflow[^1][^2]. These systems function as implants for extracting, refining, and amplifying insights already latent in your own reasoning, rather than as autonomous minds meant to replace human thought.[^21][^25] This shift addresses what Brandon Sanderson calls the core problem with AI-generated art: it steals the transformative journey of creation from the creator[^10][^11]. AI that writes your novel robs you of the growth that comes from struggling with narrative structure. AI that solves problems for you eliminates the cognitive friction needed for learning. ![Cognitive Prosthetic: Brain and Interface](/assets/blog/ai-cognitive-prosthetic/cognitive-brain-network.png) ## Abstract **AI as Cognitive Prosthetic** is a framework for treating agentic systems as extensions of human cognition rather than replacements for it. The core argument: AI should extract and amplify insight from what you already know, rather than generate conclusions on your behalf.[^10][^11][^14] The cyberpunk metaphor of "chrome," representing high-end augmentation under direct neural command, synthesizes cognitive science, human-AI collaboration research, and philosophy of mind into a single interaction model. The framework positions AI as a diagnostic interface for human intuition[^1][^2]: computational precision while maintaining human agency and preserving the cognitive friction needed for growth[^15][^32]. It solves the "outsourcing of agency" problem by treating AI like a precision surgical tool that knows your patterns and goals, reduces friction in repetitive tasks, and in this use case has no will of its own[^1][^2]. The prosthetic model emphasizes recursive refinement loops where AI presents multiple branching paths of the user's own logic, letting humans "feel out" solutions while keeping ownership of the creative process[^17][^24][^30]. It includes practical implementation patterns, diagnostic tools for identifying when AI crosses from augmentation into replacement, and architectural considerations that respect human cognitive constraints while leveraging machine computational power. **Keywords:** agentic AI, cognitive prosthetics, human-AI collaboration, augmentation, tool use, agency preservation, creative process, cognitive friction, insight extraction --- ## The Transformation Thesis What happens when we bypass the creative process entirely? Sanderson's argument centers on an insight from cognitive science and philosophy of mind: **the value of creation lies in how it changes the creator**[^10][^11][^14][^16]. ![The Transformation: Human at the Center](/assets/blog/ai-cognitive-prosthetic/human-transformation.png) When you learn to write: - Reading hundreds of books or articles expands your mental models - Struggling with plot structure develops pattern recognition - Revising failed drafts builds metacognitive awareness - Finding your voice reveals who you are The published book is evidence that this transformation occurred. AI-generated output provides the receipt without the journey[^14][^16]. > "The point of art is not the outcome, which is the receipt, but the struggle of making and how it transforms you as a person. The journey is more important than the destination"[^14][^16] The same principle applies beyond traditional art: - **Data analysis:** Manual exploration builds intuition for patterns - **Programming:** Writing code from scratch develops architectural thinking - **Problem-solving:** Iterative refinement creates mental models - **Research:** Synthesizing sources by hand deepens understanding The right amount of cognitive friction is where learning happens. AI, used well, reduces friction. When it eliminates all friction, it eliminates growth. If the goal is automation and scaling, it's another modern tool. But insight is lost in translation. --- ## The Prosthetic Alternative: Chrome, Not Oracle If AI as "answer-generator" steals the transformative journey, what's the alternative? Treat it as a cognitive prosthetic: a precision tool that extends your capabilities while keeping you in control[^1][^2][^4]. ### The Cyberpunk Metaphor In cyberpunk narratives, "chrome" refers to high-end cybernetic augmentations, implants that provide superhuman precision or strength while remaining under direct neural command[^7][^9]. Unlike autonomous robots, prosthetics have no will of their own. They extend the user's capabilities with computational precision[^1][^2]. ![Prosthetic Hand: Precision Under Control](/assets/blog/ai-cognitive-prosthetic/prosthetic-hand.png) **Key characteristics of prosthetics:** - **Precision over Autonomy:** A surgical robot provides sub-millimeter accuracy the human hand cannot achieve alone. But the surgeon controls every movement[^2] - **Feedback Loops:** High-quality prosthetics provide sensory feedback. The user "feels" resistance, pressure, and position[^27] - **Neural Integration:** The tool responds to intent, not explicit commands. It becomes an extension of thought[^11][^13] - **Preservation of Agency:** The human makes all decisions. The tool executes with enhanced capability[^1][^4] ![Cyberpunk Chrome: Human-AI Integration](/assets/blog/ai-cognitive-prosthetic/cyberpunk-hands.png) Agentic AI should work the same way: computational precision under human command, preserving human decision-making. --- ## From Generation to Extraction What if AI's real value isn't in generating answers at all? Many assume AI's value lies in generating content. But in a world of infinite, low-cost generation, the value of "answers" trends toward zero. The real value lies in **extraction**: using the agent to mine your own expertise, data, and intuition to find signal in noise[^20][^29]. ![The Journey Over Answers](/assets/blog/ai-cognitive-prosthetic/journey-over-answers.png) **The Mirror Effect:** An agentic system acts as a cognitive mirror, reflecting your logic back to you through different lenses to identify biases, gaps, and hidden assumptions[^23]. For a deeper look at what it means that LLMs mirror rather than originate psychological language, see [The Mirror Has No Face](/blog/mirror-has-no-face-ai-sounds-conscious). **Latent Knowledge Mining:** Much of human expertise is tacit. We know more than we can tell. Agentic "chrome" uses iterative prompting, Socratic questioning, and recursive summarization to surface knowledge the user didn't realize they possessed. The agent doesn't invent insights. It excavates them from the human's existing cognitive substrate[^21][^25]. **Recursive Refinement:** Unlike one-shot generation, the prosthetic model emphasizes loops. The AI presents multiple branching paths of the user's own logic. The user "feels out" which resonates, then the agent drills deeper. This mirrors how humans naturally refine ideas through dialogue and iteration[^17][^24][^30]. --- ## Critical Considerations What are the risks when even a well-designed prosthetic becomes too comfortable? ![Critical Considerations: System Awareness](/assets/blog/ai-cognitive-prosthetic/critical-considerations.png) ### Opacity and Understanding A prosthetic you don't understand can lead you astray. You're still responsible for the output, not just the tool[^16][^32]. Track the reasoning process. Understand why the AI suggests what it does. ### Dependency and Atrophy Sanderson's concern applies here: if the agent always "extracts" the insight, the human may stop developing the intuition necessary to speed the process[^10][^14][^16]. This is the cognitive equivalent of never walking because you have a wheelchair. Without training, muscles atrophy. **Prevention:** Practice core skills without AI assistance. Use the prosthetic for computational heavy-lifting, not basic cognitive operations. ### The Input/User Limit A prosthetic can only amplify what is there. If the user lacks understanding, the chrome has nothing to grip and will generate "phantom insights"[^23]. --- ## Practical Implementation: Installing the Chrome ![Implementation: The Path Forward](/assets/blog/ai-cognitive-prosthetic/implementation-path.png) To treat agentic systems as tools for insight extraction, follow these implementation steps: 1. **Define the "Seed":** Never start with a blank prompt. Provide the agent with your "raw" thoughts, half-formed ideas, or messy data. This is the "mind" from which insight will be extracted. 2. **Set the "Persona" as a Tool:** Instruct the agent to act as a specific type of prosthetic (e.g., "Act as a Socratic interlocutor designed to find the contradictions in my business plan"). 3. **Use Recursive Interrogation:** Instead of asking for a final report, ask: "Ask me five questions that will help me clarify my own thinking on this topic." 4. **Monitor the "Neural Link":** Constantly evaluate if the agent is leading you toward its answer or helping you find yours. If the former, "recalibrate the implant" by narrowing its constraints. 5. **Maintain Manual Override:** Always ensure the final synthesis and decision-making remain a biological process. The AI provides "enhanced vision," but the human decides where to look. 6. **Preserve Journey Elements:** Deliberately maintain the struggle. Use AI for computational precision, while retaining cognitive friction. Transform with the work, don't delegate the whole journey. --- ## The Audience Paradox A legitimate critique of Sanderson's thesis: he views AI art entirely from the creator's viewpoint[^14][^16]. What about the audience? > "The listeners who bought copies of that AI country tune don't give a hoot about whether the song's creator was glorified and improved by the process of creation"[^14][^16]. If the output is indistinguishable, does the process matter? ### Why the Process Still Matters ![Why Process Matters: Knowledge Web](/assets/blog/ai-cognitive-prosthetic/process-matters.png) Even from an audience perspective, process matters because: - **Depth:** Work created through struggle carries nuance and depth that pure generation lacks[^10][^11] - **Authenticity:** Audiences can often sense when work is "soulless," even if they can't articulate why[^16] - **Innovation:** Breakthroughs come from humans exploring unknown solution spaces, not from AI interpolating training data[^6][^8] - **Cultural Continuity:** If no one learns the craft, who trains the next generation? Who pushes boundaries?[^7] The prosthetic model preserves all of this. The human undergoes the transformation, develops depth, and creates in a changing world. --- ## Connected Concepts - **Extended Mind Thesis:** The philosophical idea that tools (like notebooks or AI) are literally part of our mind - **Cognitive Offloading:** Using external tools to reduce mental load - **Human-in-the-Loop (HITL):** Systems designed to require human intervention at key decision points - **Tacit Knowledge:** Knowledge difficult to transfer by writing or verbalizing what prosthetics help extract - **Neuroplasticity:** The brain's ability to reorganize itself through tool use --- ## Alternative Perspectives **The Oracle View:** Some argue AI should be an answer-engine for efficiency in non-critical tasks. Not everything needs to be a transformative journey. **The Collaborative Teammate:** Viewing AI as a peer rather than a prosthetic extension of the self[^23]. This acknowledges AI's emergent capabilities beyond pure tool status. **The Replacement View:** The belief that AI will surpass human intuition entirely, making "extraction" obsolete. Sanderson argues this misunderstands what art is[^10][^11]. **The Tool-Only View:** Rejecting the "agentic" nature of AI and treating it strictly as a static calculator. This may underutilize AI's potential for dynamic interaction. --- ## Conclusion ![Human-AI Collaboration: Two Hands Reaching](/assets/blog/ai-cognitive-prosthetic/human-ai-collaboration.png) The best decisions tend to come when you treat AI as computational precision under your own command rather than an autonomous decision-maker. Use it to extract and amplify insights from expertise you already hold, and keep enough cognitive friction that the work still builds skill and changes you. You stay the director of the process. AI is the interface. The journey matters, as does the destination. **Remember: We are the art. AI is the chrome.** This isn't anti-technology. It's pro-human. The question isn't whether to use AI. It's how to use it in a way that preserves what makes us artists, thinkers, and creators. --- ## References [^1]: Frontiers in Oral Health. (2024). AI applications in dental implantology. *Frontiers in Oral Health*, 5, 1442100. [https://doi.org/10.3389/froh.2024.1442100](https://doi.org/10.3389/froh.2024.1442100). Clinical AI-as-tool precedent where algorithms augment rather than replace practitioner judgment—the medical analog to the article's prosthetic framing. [^2]: National Center for Biotechnology Information. (2026). Surgical precision and AI integration. [https://pmc.ncbi.nlm.nih.gov/articles/PMC12786904/](https://pmc.ncbi.nlm.nih.gov/articles/PMC12786904/). Reviews AI-assisted surgical systems where the human retains control while the tool provides sub-millimeter precision cited in the chrome metaphor. [^3]: Gleecus. (2026). Agentic AI in healthcare: 2026 trends and predictions. [https://gleecus.com/blogs/agentic-ai-in-healthcare-2026-trends-predictions/](https://gleecus.com/blogs/agentic-ai-in-healthcare-2026-trends-predictions/). Industry trend piece on agentic healthcare systems; cited for the shift from passive AI oracles to workflow-integrated agents. [^4]: Kellton. (2026). Agentic AI healthcare trends 2026. [https://www.kellton.com/kellton-tech-blog/agentic-ai-healthcare-trends-2026](https://www.kellton.com/kellton-tech-blog/agentic-ai-healthcare-trends-2026). Complementary industry analysis supporting the article's agentic-AI adoption framing in healthcare contexts. [^5]: Klingemann, M. (2017). Art in the age of machine intelligence. *Arts*, 6(4), 18. [https://doi.org/10.3390/arts6040018](https://doi.org/10.3390/arts6040018). Early reflection on machine-generated art that the article contrasts with Sanderson's process-centered thesis. [^6]: Epstein, Z., et al. (2023). Art and the science of generative AI: A deeper dive. arXiv:2306.04141. [https://doi.org/10.48550/arXiv.2306.04141](https://doi.org/10.48550/arXiv.2306.04141). Interdisciplinary analysis of generative AI's cultural impact, supporting the article's discussion of innovation and audience reception. [^7]: Cyberpunk Wiki. (n.d.). Implants. [https://cyberpunk.fandom.com/wiki/Implants](https://cyberpunk.fandom.com/wiki/Implants). Reference for the cyberpunk "chrome" metaphor of high-end cybernetic augmentations under direct user control. [^8]: GameSpot. (2021). How close are we to Cyberpunk 2077's cyberware augmentations? [https://www.gamespot.com/articles/how-close-are-we-to-cyberpunk-2077s-cyberware-augmentations-in-real-life/1100-6517986/](https://www.gamespot.com/articles/how-close-are-we-to-cyberpunk-2077s-cyberware-augmentations-in-real-life/1100-6517986/). Popular-science context for real-world prosthetic and BCI technology that grounds the article's cyberpunk metaphor. [^9]: Voidline. (2026). What is this cyberpunk style in 2026? [https://voidline.co.uk/blogs/news/what-is-this-cyberpunk-style-in-2026](https://voidline.co.uk/blogs/news/what-is-this-cyberpunk-style-in-2026). Cultural commentary on cyberpunk aesthetics cited for the article's chrome-as-augmentation framing. [^10]: Sanderson, B. (2026). We are the art [Keynote speech]. YouTube. [https://www.youtube.com/watch?v=mb3uK-_QkOo](https://www.youtube.com/watch?v=mb3uK-_QkOo). Primary source for Sanderson's thesis that AI-generated output steals the transformative journey of creation—the article's central argument. [^11]: Sanderson, B. (2026). The hidden cost of AI art: Brandon Sanderson's keynote. *Brandon Sanderson Blog*. [https://www.brandonsanderson.com/blogs/blog/ai-art-brandon-sanderson-keynote](https://www.brandonsanderson.com/blogs/blog/ai-art-brandon-sanderson-keynote). Written companion to the keynote elaborating how creation changes the creator and why the process, not the product, is art's primary value. [^12]: PC Gamer. (n.d.). What Cyberpunk-style implant would you want in real life? [https://www.pcgamer.com/what-cyberpunk-style-implant-would-you-want-in-real-life/](https://www.pcgamer.com/what-cyberpunk-style-implant-would-you-want-in-real-life/). Illustrative popular discussion of augmentation preferences; supports the article's prosthetic-as-extension metaphor in gaming culture. [^13]: Open Data Science. (2026). Agentic AI skills 2026. [https://opendatascience.com/agentic-ai-skills-2026/](https://opendatascience.com/agentic-ai-skills-2026/). Skills-oriented overview of agentic AI capabilities cited in the article's implementation context. [^14]: The Silver Key. (2026). Brandon Sanderson articulates the exact problem with AI in the arts. [https://thesilverkey.blogspot.com/2026/02/brandon-sanderson-articulates-exact.html](https://thesilverkey.blogspot.com/2026/02/brandon-sanderson-articulates-exact.html). Commentary distilling Sanderson's "receipt without the journey" argument that the article adopts as its core framing. [^15]: Lakoff, G., & Johnson, M. (1980). *Metaphors We Live By*. University of Chicago Press. [https://doi.org/10.7208/chicago/9780226470993.001.0001](https://doi.org/10.7208/chicago/9780226470993.001.0001). Foundational cognitive-linguistics text on conceptual metaphor—the theoretical basis for the article's prosthetic and chrome framing. [^16]: Sanderson, B. (2026). We are the art [Keynote speech]. YouTube. [https://www.youtube.com/watch?v=mb3uK-_QkOo](https://www.youtube.com/watch?v=mb3uK-_QkOo). Sanderson's audience-side critique—that listeners may not care about the creator's journey—which the article engages as a legitimate counterargument. [^17]: Wickelgren, W. A. (1974). *How to Solve Problems*. W.H. Freeman. Problem-solving methodology supporting the article's recursive refinement loops and branching-path exploration. [^18]: De Bono, E. (1985). *Six Thinking Hats*. Little, Brown. Structured multi-perspective thinking framework paralleling the article's branching-path prosthetic model. [^19]: Bransford, J. D., Brown, A. L., & Cocking, M. R. (2000). *How People Learn*. National Academy Press. [https://doi.org/10.17226/9853](https://doi.org/10.17226/9853). National Academies synthesis on learning through active struggle—the cognitive-science basis for preserving friction in the prosthetic model. [^20]: Tausczik, Y. R., & Pennebaker, J. W. (2010). The psychological meaning of words: LIWC and computerized text analysis methods. *Journal of Language and Social Psychology*, 29(1), 24-54. [https://doi.org/10.1177/0261927X09351676](https://doi.org/10.1177/0261927X09351676). Text-analysis methods for surfacing latent patterns in user language—the computational basis for insight extraction. [^21]: Goldvarg, E., & Johnson-Laird, P. N. (2001). Naive causality: A mental model theory of causal meaning and reasoning. *Cognitive Science*, 25(4), 565-610. [https://doi.org/10.1207/s15516709cog2504_2](https://doi.org/10.1207/s15516709cog2504_2). Mental-model theory supporting the article's claim that agents can excavate tacit causal knowledge users already possess. [^22]: Screen Rant. (2025). Brandon Sanderson weighs in on ongoing AI debate. [https://screenrant.com/brandon-sanderson-author-responds-ai-debate-art/](https://screenrant.com/brandon-sanderson-author-responds-ai-debate-art/). Press coverage of Sanderson's AI-in-arts position cited for broader cultural context. [^23]: Literature and Latte Forum. (2026). Brandon Sanderson: What it means to be human | Art in the AI era. [https://forum.literatureandlatte.com/t/brandon-sanderson-what-it-means-to-be-human-art-in-the-ai-era/152972](https://forum.literatureandlatte.com/t/brandon-sanderson-what-it-means-to-be-human-art-in-the-ai-era/152972). Community discussion of Sanderson's human-agency arguments and the "cognitive mirror" framing in the article. [^24]: Wang, X., Wei, J., Schuurmans, D., Le, Q., Chi, E., Narang, S., Chowdhery, A., & Zhou, D. (2023). Self-consistency improves chain of thought reasoning in language models. In *International Conference on Learning Representations* (ICLR 2023). arXiv:2203.11171. [https://doi.org/10.48550/arXiv.2203.11171](https://doi.org/10.48550/arXiv.2203.11171). Multiple-path reasoning with consistency voting—the algorithmic analog to the article's recursive refinement loops. [^25]: Let's Data Science. (2026). Sanderson frames art as personal transformation. [https://letsdatascience.com/news/sanderson-frames-art-as-personal-transformation-61299092](https://letsdatascience.com/news/sanderson-frames-art-as-personal-transformation-61299092). Secondary summary of Sanderson's transformation thesis supporting the article's insight-extraction framing. [^26]: Park, J. S., O'Brien, J., Cai, C. J., Morris, M. R., Liang, P., & Bernstein, M. S. (2023). Generative agents: Interactive simulacra of human behavior. In *UIST '23*. arXiv:2304.03442. [https://doi.org/10.48550/arXiv.2304.03442](https://doi.org/10.48550/arXiv.2304.03442). Agent architecture research cited for emergent agentic capabilities that the article distinguishes from the prosthetic model. [^27]: Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., & Qin, Y. (2004). An integrated theory of the mind. *Psychological Review*, 111(4), 1036-1060. [https://doi.org/10.1037/0033-295X.111.4.1036](https://doi.org/10.1037/0033-295X.111.4.1036). ACT-R cognitive architecture supporting the article's neural-integration and sensory-feedback prosthetic properties. [^28]: Akitra. (2026). Developing standards for agentic AI. [https://akitra.com/blog/developing-standards-for-agentic-ai/](https://akitra.com/blog/developing-standards-for-agentic-ai/). Industry standards discussion for agentic AI governance cited in the article's implementation considerations. [^29]: Pennebaker, J. W., Boyd, R. L., Jordan, K., & Blackburn, K. (2015). The development and psychometric properties of LIWC2015. University of Texas at Austin. [https://doi.org/10.15781/T29G6Z](https://doi.org/10.15781/T29G6Z). Updated LIWC instrument for computerized text analysis supporting the article's insight-mining methods. [^30]: Yao, S., Yu, D., Zhao, J., Shafran, I., Griffiths, T. L., Cao, Y., & Narasimhan, K. (2023). Tree of thoughts: Deliberate problem solving with large language models. In *Advances in Neural Information Processing Systems 36* (NeurIPS 2023). arXiv:2305.10601. [https://doi.org/10.48550/arXiv.2305.10601](https://doi.org/10.48550/arXiv.2305.10601). Branching-path exploration algorithm paralleling the article's prosthetic refinement loops where users choose among paths of their own logic. [^31]: Brown, A. L., & Day, J. D. (1983). Macrorules for summarizing texts: The development of expertise. *Journal of Verbal Learning and Verbal Behavior*, 22(1), 1-14. [https://doi.org/10.1016/S0022-5371(83)80002-4](https://doi.org/10.1016/S0022-5371(83)80002-4). Expert summarization strategies supporting iterative refinement and recursive summarization in the prosthetic workflow. [^32]: Sweller, J., Ayres, P., & Kalyuga, S. (2011). *Cognitive Load Theory*. Springer. [https://doi.org/10.1007/978-1-4419-8126-4](https://doi.org/10.1007/978-1-4419-8126-4). Cognitive load theory grounding the article's emphasis on preserving productive friction and respecting human cognitive constraints. --- ## License This work is licensed under a [Creative Commons Attribution 4.0 International License](http://creativecommons.org/licenses/by/4.0/).