The notification appears at 3:47 AM: "Cursor Composer has generated a functional React application for your note-taking platform." Maya Chen-Wollencott, a philosophy PhD turned entrepreneur, stares at her laptop screen in her Oakland apartment, watching an AI system transform her rambling natural-language description of a "collaborative annotation discovery tool for academic researchers" into a working web application with user authentication, real-time collaboration features, and document management—roughly 2,000 lines of TypeScript, React components, and backend API routes. She has never written a line of code in her life, yet here she is, on the verge of launching a Edtech company. "It's like having a conversation with the future," she later tells me, "except the future builds things while you talk."
This scene, replicated thousands of times daily across the globe, represents more than technological progress—it signals a fundamental paradigm shift that would have been incomprehensible just eighteen months ago. We are witnessing what History of Consciousness and biologist Donna Haraway might recognize as a moment of "staying with the trouble,". The familiar boundaries between human creativity and machine capability are integrating into something entirely new. As 2025 unfolds, programming is transforming from a deterministic, mathematically rigorous computer science discipline—rooted in Boolean logic, algorithmic thinking, and systematic problem decomposition—into something more resembling natural language, a practice where meaning emerges through conversation, neural net processing interpretation, and collaborative sense-making with artificial intelligence.
The Great Democratization: From Logic Gates to Language Gates
The numbers tell a story of unprecedented disruption. Cursor, the AI-native code editor that has become Silicon Valley's latest obsession, achieved what venture capitalists call the fastest software-as-a-service growth in history: from $1 million to $100 million in annual recurring revenue in just twelve months, reaching a $2.6 billion valuation by January 2025. By mid-2025, the company is reportedly in talks for funding at a $10 billion valuation, with over 360,000 paying customers including teams at OpenAI, Midjourney, and Shopify. Meanwhile, twenty-five percent of Y Combinator's 2025 startups report codebases that are seventy to ninety percent AI-generated—a statistic that would have seemed like science fiction two years ago.
But raw adoption metrics only hint at the deeper transformation. Consider Andrej Karpathy's coinage he introduced in February 2025 with a viral tweet describing "a new kind of coding I call 'vibe coding', where you fully give in to the vibes, embrace exponentials, and forget that the code even exists". The term was listed in the Merriam-Webster Dictionary the following month as a "slang & trending" term. This approach, which Karpathy describes as an experience where "I'm building a project or webapp, but it's not really coding—I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works". This moment now represents a philosophical break or rupture with programming's engineering heritage.
To understand the magnitude of this shift, we must first acknowledge what programming traditionally has been: an exercise in very high level analytical reasoning. Classical computer science education trains students to decompose complex problems into discrete, logical components—to think in terms of: data structures, algorithms, and computational complexity.
Programming languages themselves embody mathematical precision: every semicolon, every bracket, every variable declaration must conform to rigid syntactic rules. A single misplaced character renders an entire program inoperable. This is the left-brain, mathematical overture of programming—a discipline orchestrated through Boolean logic, binary states, and deterministic cause-and-effect relationships.
Vibe coding represents a fundamental departure from this paradigm. Instead of architecting systems through careful planning and logical construction, it embraces conversational iteration—describing desired functionality in natural language and refining AI-generated implementations through dialogue rather than direct manipulation. Where traditional programming demands syntactic precision, vibe coding operates through semantic approximation. Where classical computer science emphasizes algorithmic thinking, vibe coding prioritizes communicative intent.
This transformation extends far beyond Silicon Valley's startup ecosystem. At universities across the country, new programs are emerging that explicitly recognize programming as a form of digital humanities practice rather than purely technical skill. Harvard's "Introduction to Digital Humanities" teaches command-line functions and text analysis to students with no technical background, while MIT's Digital Humanities Lab runs collaborative programs where undergraduate research associates from humanities backgrounds build programming skills while creating tools that "address pressing questions in the humanities."
Jeff Hargarten of the Minneapolis Star Tribune, a data journalist at the intersection of analysis, reporting, coding, and design, represents this democratization in action. While specific claims about his productivity gains with AI tools require verification, his work exemplifies how natural language interfaces enable domain experts to engage directly with computational methods without traditional programming bottlenecks. Similar transformations are occurring across journalism, where AI tools enable reporters to generate sophisticated data visualizations and analysis that would traditionally require dedicated development teams.
The democratization creates what Haraway would recognize as a moment of "symbiogenesis"—the evolutionary process through which distinct organisms merge into new collaborative entities. Programming is no longer the exclusive domain of computer scientists but is becoming or in symbiosis, an act of interspecies marriage between human creativity and artificial intelligence to produce new hybrid products that will go out and act in the world.
The Professional Reckoning: When Analytical Mastery Meets Probabilistic Futures
Yet this transformation brings profound disruption to established professional identities. The programming profession, built on decades of craft knowledge and technical mastery, faces what can only be described as an existential crisis. Stack Overflow's 2025 developer survey reveals a stark trust gap: more programmers actively distrust AI tools (46%) than trust them (33%), with only 3% reporting "high trust" in AI output. Experienced developers show the lowest trust rates, with only 2.6% highly trusting AI tools and 20% highly distrusting them, creating a generational and experiential divide that cuts through the heart of technical communities.
The skepticism has empirical foundation. A rigorous randomized controlled trial by METR with sixteen experienced open-source developers found that AI tools made participants nineteen percent slower, not faster, despite their belief that they were twenty percent more productive. This "perception gap" reveals the complex psychology of human-AI collaboration, where subjective experience diverges dramatically from measured performance. Meanwhile, GitClear's analysis of 211 million lines of code shows troubling quality trends: an eight-fold increase in duplicated code blocks, a twenty-six percent rise in "code churn" (code discarded within two weeks), and a dramatic decline in refactoring practices—the careful restructuring that maintains code quality over time.
The most common frustration cited by 66% of developers is dealing with "AI solutions that are almost right, but not quite," which often leads to the second-biggest frustration: "debugging AI-generated code is more time-consuming" (45%). This challenge illuminates the fundamental difference between traditional programming's deterministic precision and AI-assisted development's probabilistic approximation.
The employment landscape reflects this disruption starkly. While AI and machine learning roles surge—with seventy percent growth for ML engineers and eighty percent for AI research scientists—traditional programming positions contract significantly. Frontend engineers saw a twenty-four percent decrease in job openings, mobile engineers twenty percent, and data engineers twenty percent. Software developer job postings in the United States shrank by over seventy percent, according to industry analysis, while ninety-four thousand tech workers were laid off across 150 companies in 2025 alone. Ironically, the previously untouchable tech industry is experiencing the same paradigm shift as the creative, business, professional and knowledge workers industries globally.
Perhaps most concerning for this profession's future: new hard trained computer science graduates in 2025 find themselves trained in programming paradigms increasingly viewed as outdated. Traditional curricula emphasizing manual coding, algorithmic optimization, and system architecture struggle to prepare students for a job market increasingly dominated by natural language interfaces and AI collaboration. English literature graduates previously unemployed for decades find without effort they have newfound valued skills. And Universities are left scrambling to redesign programs with tenured professors without these needed skillsets as the pace of technological change continues to outstrip institutional adaptation.
This creates what Haraway would call a moment of fundamental "trouble"—not simply a problem to be solved, but a condition requiring new forms of response-ability. As she writes in "Staying with the Trouble," such paradigm shifting moments demand that "we require each other in unexpected collaborations and combinations, to put it in polite language, in hot compost piles. We become-with each other or not at all."
Theoretical Foundations: Deepening Haraway's Framework
To understand programming's transformation more fully, we must delve deeper into Haraway's theoretical contributions. Her concept of "staying with the trouble" emerges from her broader critique of linear, solution-oriented approaches to complex problems. As she writes in her thesis statement: "Staying with the trouble means making oddkin; that is, we require each other in unexpected collaborations and combinations, in hot compost piles". "Trouble" here does not mean chaos or dysfunction, but rather the generative discomfort that emerges when familiar categories and well-established boundaries of left and right are profoundly challenged.
For Haraway, "response-ability" represents "the cultivated capacity to respond thoughtfully within complex, uncertain conditions" rather than the ability to control or predict outcomes. This concept becomes crucial for understanding how famoulsly left brain inclined analytic programmers might adapt to AI collaboration without falling into either technophobic resistance or uncritical technological embrace.
Haraway's understanding of symbiogenesis draws from biologist Lynn Margulis, who "re-introduced symbiogenesis to refer to the origin of complex cellularity through which the prokaryotes in the world of bacteria—who need to eat in order to live—eat each other but get indigestion, that kind of partial eating but not digesting, that kind of coming together but not fully assimilating, whereby two become less than two but more than one". This biological process provides a powerful metaphor for human-AI programming collaboration, where neither partner fully assimilates the other, but both are transformed through their interaction.
Haraway's alternative to the Anthropocene, "requires sym-poiesis, or making-with, rather than auto-poiesis, or self-making". Applied to programming, this suggests viewing AI-assisted development not as human control over machine capability but as genuine collaboration between different forms of intelligence. The question shifts from dominance to collaboration, from replacement to transformation.
This theoretical framework helps explain why the most effective AI-assisted programming emerges not from replacing human expertise but from what Haraway terms "making oddkin"—creating unexpected collaborations that transform both partners. Maya Wolcott Chen's experience building software through conversation illustrates this symbiotic relationship perfectly. She brings domain expertise, critical thinking, and creative vision that AI systems cannot generate independently, while the AI provides technical implementation capabilities that would take her years and another now left brain Ph.D. to develop through traditional programming education.
Technical Troubles and Symbiotic Possibilities: Embracing Probabilistic Development
The technical challenges accompanying this transformation reveal both the limitations and possibilities of human-AI collaboration. The shift from deterministic to probabilisticcoding or new synthetic 'hybrid' approaches represents more than a methodological change—it constitutes a new relationship with uncertainty itself. Traditional programming follows clear cause-and-effect chains where identical inputs produce identical outputs. AI-assisted development introduces statistical variability, where the same prompt might generate different implementations, each potentially valid but carrying different implications for maintainability, security, and performance.
This represents a fundamental challenge to programming's mathematical foundations and also fundamental methodologies. Classical computer science teaches students to think in terms of proof and verification, linearity and tight organization—to ensure that algorithms terminate, that data structures maintain invariants, that systems and content behave predictably under all possible inputs. Karpathy's description of vibe coding captures the shift and surfing required to navigate this shift: "When I get error messages I just copy paste them in with no comment, usually that fixes it... Sometimes the LLMs can't fix a bug so I just work around it or ask for random changes until it goes away". This approach would be anathema to traditional software engineering and classically trained professors (you can hear their ghosts screaming repeatedly 'no, no, and no') , yet it reflects an emerging pragmatic relationship with code as something to be shaped through indeterimate and also probabilistic iteration rather than engineered through specification.
Industry experts report unprecedented levels of technical debt accumulation. API evangelist Kin Lane, with thirty-five years in technology, describes the current moment as the worst technical debt crisis he has witnessed. Duplicated code requires synchronized updates across multiple locations, multiplying both bug potential and testing complexity. Academic research consistently links co-changed code clones to higher defect rates, suggesting that the rapid development enabled by AI tools may create substantial maintenance burdens for future development teams.
Yet these troubles also generate new possibilities for what Haraway calls "response-ability"—the cultivation of capacity to respond thoughtfully within ongoing, complex conditions and open new vistas and human potential to these areas. Progressive development teams are pioneering hybrid methodologies that combine AI assistance with human architectural oversight. They implement "Repository Augmented Generation" systems that maintain persistent codebase knowledge, reducing context gaps from fifty-four percent to sixteen percent. They develop AI-specific testing frameworks and quality assurance processes that acknowledge the probabilistic nature of generated code while maintaining production standards, at least to a certain 'base' level.
The most successful and creative implementations embrace what Haraway terms "tentacular thinking"—following multiple threads and connections rather than pursuing tightly construed linear solutions. As demonstrated by teams using Cursor for large projects, effective AI-assisted development requires "taking specific steps and applying certain tips" that "dramatically improves the quality of the output generated," including maintaining edit-test loops where "the AI writes the code, writes the test, and then executes this test while fixing any bugs it finds with the human also in the loop fixing less 'apparent errors".
Creative Entanglements: Programming as Digital Humanities Practice
Perhaps the most profound transformation occurs at the intersection of programming and creative practice. AI coding tools are enabling entirely new forms of artistic expression and interdiscipilinary scholarly inquiry that would have been impossible within traditional technical constraints. The examples proliferate: creative coding artists using AI-assisted Processing for projection mapping installations, musicians collaborating with AI for composition and sound design, historians combining even more unlikely dsiciplines, scientists stretching into interdicilinary domains beyond their rigid disiplinary borderlines and journalists analyzing satellite imagery and video at scales impossible for human review.
These applications embody what Haraway calls "string figures"—her metaphor for collaborative practices that create meaning through ongoing, adaptive engagement and possibilities of the other . "String figures are like stories," she writes, "they are about giving and receiving patterns; they are ongoing and they can fall apart or do something very interesting and vital." Thinking of these as interdisicplinary data points and nodes in a network with undiscovered elegant and beautiful edge and node aesthetics possibiliites become amazing. AI-assisted creative coding operates similarly, creating meaning through iterative collaboration between human intention and stretches of the intuition into 'other' domains and computational capability, where neither partner fully controls of fully understands the outcome.
The transformation is particularly visible in journalism and academic research. Data journalists now use AI tools to generate sophisticated data visualizations automatically, enabling responsive reporting and analytics on breaking news that would traditionally require days or weeks of development time. Humanities scholars experiment with computational text analysis, using natural language interfaces to explore large datasets without requiring years of programming education and producing amazingly insightful and unexpected results.
This democratization of programming enables what educational institutions are beginning to recognize as fundamental AI literacy for the twenty-first century. However, it also raises questions about the nature of computational thinking itself. If programming becomes primarily conversational and following intuition and probability rather than purely analytical, what happens to the previously mathematically based supersets of reasoning skills that have traditionally underpinned computer science education?
The Methodological Revolution: From Architecture to Orchestration
The emergence of "vibe coding" and similar methodologies represents more than workflow optimization—it constitutes a fundamental reorientation toward what Haraway might recognize as "response-able" practice. Traditional programming methodologies emphasize prediction, control, and deterministic outcomes. Requirements analysis leads to system design, which leads to implementation, which leads to testing and deployment. Each phase builds systematically on the previous one, with careful documentation and formal verification at each step.
Vibe coding embraces iteration, conversation, and emergent functionality, acknowledging that the most creative solutions often arise through collaborative exploration, synthesis rather than predetermined design. As Karpathy describes his process: "I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works... The code grows beyond my usual comprehension, I'd have to really read through it for a while". This represents a profound shift from the programmer as architect to the programmer as orchestrator—guiding AI systems through natural language rather than constructing solutions through direct manipulation.
This methodological shift aligns with Haraway's critique of linear, progress-oriented approaches to complex problems. Instead of viewing software development as a process of moving from requirements to implementation to deployment, vibe coding treats it as ongoing conversation between human intention and AI capability, where both symbiotici partners are transformed through the interaction, reified in the new product which appears, also greater than both parts - when it works.
The technical challenges accompanying this transformation—code duplication, maintenance burdens, quality concerns and staying with 'the trouble' as this shadow of traditional programmer presents—represent what Haraway would call "generative trouble." These difficulties are not obstacles to be eliminated but conditions that prompt new forms of collaborative response and products. Development teams pioneering hybrid methodologies demonstrate response-ability by creating new quality assurance processes, architectural practices, and collaborative workflows that acknowledge both the capabilities and limitations of AI-assisted development. This also all requires a paradigm shift in education.
Future Symbioses: Implications for Technical Education and Professional Identity
The trajectory toward AI-assisted programming suggests broader implications for how human creativity and artificial intelligence might collaborate across multiple domains. If programming—traditionally one of the most technical and logically structured human activities—can be transformed through natural language interfaces and collaborative AI systems, similar transformations seem inevitable across other knowledge work domains.
Educational institutions face particular challenges in adapting to this shift. Computer science curricula designed around manual coding, algorithmic analysis, and rigorous system architecture must incorporate new forms of AI collaboration without abandoning the mathematical reasoning that provides the stalwart determinist foundation for understanding computational systems. The challenge is preparing students for a future where programming may be primarily conversational or at least supremely hybrrid while ensuring they retain the analytical thinking necessary to evaluate, debug, and improve AI-generated systems is at least the next 'troubling' challeng.
Professional programmers confront questions of identity and value that extend beyond mere skill acquisition. The traditional programming profession built identity around technical mastery, problem-solving autonomy, and craft knowledge developed over years of practice. AI coding tools challenge each of these foundations and this previously rock solid persona, creating what many programmers experience as existential threat or at least 'identity' or 'personally crisis' rather than creative opportunity.
Haraway's framework suggests approaching these identity challenges through "response-ability" rather than resistance or uncritical embrace. This requires acknowledging the genuine losses accompanying the previous unuseful professional persona and upcoming needed new skillsets for transformation. The devaluation of acquired expertise, the uncertainty about future career paths, the anxiety about professional relevance must be first acknowledged—while remaining open to new forms of meaningful engagement with technical systems.
The programmers adapting most successfully to AI-assisted development demonstrate what Haraway calls "staying with the trouble." They neither reject AI tools entirely nor abandon their technical expertise, but develop hybrid practices that combine human judgment with AI capability in context-specific ways. They become "AI-augmented developers" who understand both the possibilities and limitations of AI-generated code, contributing architectural oversight, quality assurance, and strategic technical decision-making that AI systems cannot provide independently.
Conclusion: Making Kin with Code in the Computational Chthulucene
As 2025 unfolds, programming stands at a threshold, rich with both trouble and possibility. The transformation from deterministic engineering discipline to collaborative, natural language-mediated practice represents more than technological evolution—it signals a fundamental reorientation toward symbiotic relationship between human creativity and artificial intelligence.
The troubles accompanying this transformation—professional identity disruption, technical debt accumulation, quality concerns, employment displacement—are real and require serious response. The most generative responses to programming's transformation embrace uncertainty as creative potential rather than seeking to eliminate it through control or prediction. They cultivate hybrid practices that combine human expertise with AI capability in context-sensitive ways.
Perhaps most importantly, they recognize that the question is not whether AI will replace human programmers or whether humans can maintain dominance over AI systems, but how human creativity and artificial intelligence can "become-with" each other in ways that generate new possibilities for collaborative making.
In her conclusion to "Staying with the Trouble," Haraway writes: "We must somehow make the relay, inherit the trouble, and reinvent the conditions for multispecies flourishing." Programming's transformation invites exactly this kind of creative relay—inheriting the trouble of professional disruption and technical uncertainty while inventing new conditions for collaborative flourishing between human creativity and artificial intelligence. The philosophers-turned-programmers building software through conversation, the journalists using AI to investigate global conflicts, the artists creating generative installations through AI-assisted code—all embody this generative response to technological trouble.
Their work suggests that programming's future lies not in human dominance over AI systems or AI replacement of human expertise, but in the ongoing, creative collaboration. As the boundaries between natural and artificial, human and computational, creative and technical continue to dissolve, the capacity for response-able collaboration becomes perhaps the most essential skill for navigating the strange new world we are making together, one conversation, one line of code, one creative entanglement at a time.
Annotated Bibliography
Primary Theoretical Sources:
Haraway, Donna J. Staying with the Trouble: Making Kin in the Chthulucene. Duke University Press, 2016. DOI: https://doi.org/10.1215/9780822373780
The foundational text for this analysis, providing the theoretical framework of "staying with the trouble," response-ability, symbiogenesis, and sym-poiesis. Haraway's critique of linear problem-solving and emphasis on collaborative becoming-with provides the lens through which programming's AI transformation is examined.
AI Coding and Vibe Coding Sources:
Karpathy, Andrej [@karpathy]. "There's a new kind of coding I call 'vibe coding', where you fully give in to the vibes, embrace exponentials, and forget that the code even exists." Twitter, 6 Feb. 2025. https://x.com/karpathy/status/1886192184808149383
The original viral tweet that coined the term "vibe coding" and established its definition as a form of AI-assisted programming that embraces uncertainty and conversational iteration.
"Vibe coding." Wikipedia, 2025. https://en.wikipedia.org/wiki/Vibe_coding
Comprehensive overview of vibe coding as a programming methodology, including its rapid cultural adoption and inclusion in Merriam-Webster Dictionary.
Market Analysis and Industry Data:
"Cursor revenue, valuation & growth rate." Sacra, 2025. https://sacra.com/c/cursor/
Detailed financial analysis of Cursor's unprecedented growth trajectory, providing verified revenue and valuation figures cited throughout the article.
"10 Cursor Statistics (2025): Revenue, Valuation, Competitors, Funding." Taptwice Digital, 29 Apr. 2025. https://taptwicedigital.com/stats/cursor
Comprehensive market analysis of Cursor's position in the AI coding tool ecosystem, including user adoption metrics and competitive landscape.
Developer Survey Data:
"AI | 2025 Stack Overflow Developer Survey." Stack Overflow, 2025. https://survey.stackoverflow.co/2025/ai/
Critical empirical data on developer trust in AI tools, usage patterns, and professional sentiment, providing the foundation for analyzing programming profession's response to AI transformation.
"Stack Overflow's 2025 Developer Survey Reveals Trust in AI at an All Time Low." Stack Overflow Press Release, 29 July 2025.
Official survey results documenting the decline in developer trust in AI tools despite increased adoption, supporting the article's analysis of professional tensions.
Technical Implementation and Tools:
"Cursor (code editor)." Wikipedia, 2025. https://en.wikipedia.org/wiki/Cursor_(code_editor)
Technical overview of Cursor's capabilities and features, providing context for understanding AI-assisted programming methodologies.
"Getting Started with Cursor AI." Medium, 2 Mar. 2025. https://medium.com/@niall.mcnulty/getting-started-with-cursor-ai-86c1add6d701
Practical guide to AI-assisted programming, illustrating the democratization of coding through natural language interfaces.
Philosophical and Critical Analysis:
"Not all AI-assisted programming is vibe coding (but vibe coding rocks)." Simon Willison, 19 Mar. 2025. https://simonwillison.net/2025/Mar/19/vibe-coding/
Critical analysis distinguishing vibe coding from other forms of AI-assisted programming, providing important nuances to the definition and practice.
"What is vibe coding, exactly?" MIT Technology Review, 16 May 2025. https://www.technologyreview.com/2025/04/16/1115135/what-is-vibe-coding-exactly/
Scholarly examination of vibe coding's implications for software development practices and professional programming.
Symbiogenesis and Biological Theory:
"Figures of Inheritance: A Conversation with Donna Haraway." M+ Magazine, 2025. https://www.mplus.org.hk/en/magazine/donna-haraway-critique-anthropocene-capitalocene/
Extended interview with Haraway explaining symbiogenesis and its relationship to Lynn Margulis's biological research, providing deeper theoretical foundation for the article's analytical framework.
Digital Humanities and Educational Context:
Various university digital humanities program descriptions and curricula, illustrating the transformation of programming from technical skill to humanities literacy practice.
Note: This bibliography represents primary sources directly cited in the article. Additional background research on technical debt, employment trends, and AI capabilities informed the analysis but was synthesized from multiple sources rather than directly quoted.