South American Publishing books@southam.pub 20261106T162924Z eng

9789915996639-en-1 03 01 15 9789915996639 06 10.62486/978-9915-9966-3-9 00 ED E107 00 01 01 Superintelligence Series Volume 4: One Person Company: Artificial Intelligence, Autonomous Entrepreneurship and the Future of Business 1 B01 21 0000-0003-0425-7201 Carlos Alberto Gómez Cano Carlos Alberto Gómez Cano 1 01 eng 10 BUS025000 BUSINESS & ECONOMICS / Entrepreneurship 10 BUS115000 BUSINESS & ECONOMICS / Freelance & Self-Employment 10 COM100000 COMPUTERS / Artificial Intelligence / Generative AI 10 BUS041000 BUSINESS & ECONOMICS / Management 93 UYQE 20 Artificial intelligence; One-person companies; Entrepreneurship; Solo self-employment; Multi-agent systems; Platform dependence 30 00 The possibility of an enterprise operated by a single founder with extensive support from artificial intelligence raises questions about business formation, the organization of work, and the distribution of economic responsibility. Bringing together thirteen chapters, this volume examines the empirical basis and theoretical limits of the one-person company. Studies of business applications, nonemployer establishments, and European self-employment patterns find no consistent evidence of a broad entrepreneurial surge attributable to AI diffusion. Analyses of open models, research authorship, and software repositories show that individual contributions remain possible, although prominent technical resources often depend on organizations or concentrated groups of maintainers. Reviews of experimental productivity research identify gains in the speed of selected tasks, with more variable effects on quality and creativity. Research on smaller enterprises and venture creation points to managerial capability, founder judgment, and the fit between tools and tasks as conditions shaping adoption and outcomes. The volume also considers how theories of the firm can account for AI-assisted production, what multi-agent systems can accomplish in bounded workflows, and why legal personality, liability, taxation, and platform dependence remain unresolved or consequential. The prospect of a one-person unicorn is treated as an unverified hypothesis rather than an observed business model. Across these perspectives, the central distinction is between automating activities and sustaining an accountable enterprise. Artificial intelligence may expand what an individual can accomplish, but the viability and independence of a firm continue to depend on institutional arrangements, oversight, and the distribution of risk. 04 00 Chapter 1: Business formation and the public release of generative artificial intelligence: an interrupted time-series analysis; Chapter 2: Nonemployer business growth in artificial-intelligence-exposed industries of the United States; Chapter 3: Enterprise artificial intelligence adoption and solo self-employment in knowledge-intensive activities: an ecological study across European countries; Chapter 4: Individual and organizational creators of the most-downloaded open artificial intelligence models; Chapter 5: Single-authored papers in artificial intelligence: a time-trend analysis of arXiv metadata; Chapter 6: Contributor concentration in popular open-source artificial-intelligence agent frameworks; Chapter 7: Generative artificial intelligence and individual worker productivity: a systematic review of experimental evidence; Chapter 8: Determinants of artificial intelligence adoption in micro and small enterprises: a systematic review; Chapter 9: Generative artificial intelligence in entrepreneurship: a systematic review of empirical studies on opportunity recognition and venture creation; Chapter 10: From superstar firms to super-individuals: theoretical foundations of the one-person company; Chapter 11: Multi-agent artificial intelligence systems as the workforce of the one-person company; Chapter 12: DOI: 10.62486/978-9915-9966-3-9.Ch12 Legal personality, liability and taxation of artificial-intelligence-run one-person companies; Chapter 13: The one-person unicorn hypothesis: systemic risks, platform dependence and future scenarios 01 00 03 02 https://southam.pub/book-catalog/2026/978-9915-9966-3-9/978-9915-9966-3-9_Portada.png 1 03 06 10.62486/978-9915-9966-3-9.Ch01 01 04 1 Business formation and the public release of generative artificial intelligence: an interrupted time-series analysis 1 A01 21 0000-0002-9455-4324 Libardo Ramón Polanía Libardo Ramón Polanía 2 A01 21 0009-0008-6878-0441 Rómulo Esteban Montilla Rómulo Esteban Montilla 01 eng 30 00 Introduction: The public release of generative artificial intelligence in November 2022 promised to lower the cost of starting and running a firm, yet its effect on aggregate business formation remains unmeasured. Methods: We used monthly business applications from the US Census Bureau Business Formation Statistics, seasonally adjusted, from January 2018 through August 2026, and fitted an interrupted time-series model with a break at November 2022, estimating level and trend changes with Newey-West standard errors (12 lags) for the aggregate and for three knowledge-intensive sectors. Results: At the national level the immediate shift was −49,832 applications (p = 0.099) and the monthly trend slowed by 1,602 applications (p = 0.034), from a pre-release slope of 3,631. Sectoral responses diverged, with Information falling 2,123 applications at the break (p = 0.032) and Management of companies accelerating by 44 applications per month (p < 0.001). Conclusions: The release of generative artificial intelligence coincided with a slowdown in aggregate business formation and with divergent, sector-specific responses, a pattern consistent with skill-biased reallocation rather than a uniform startup surge. 2 03 06 10.62486/978-9915-9966-3-9.Ch02 01 04 2 Nonemployer business growth in artificial-intelligence-exposed industries of the United States 1 A01 21 0009-0002-0970-2810 Edixon Chacón Guerrero Edixon Chacón Guerrero 01 eng 30 00 Introduction: Whether generative artificial intelligence enables the one-person company or instead concentrates advantage in larger employers is an open empirical question. Methods: We used national nonemployer establishment counts from the US Census Bureau Nonemployer Statistics, 2018 to 2024, aggregated to two-digit NAICS sectors, and compared compound annual growth between AI-exposed sectors (Information, Finance and insurance, Professional services) and the remaining sectors. Results: Nonemployer establishments rose from 26.5 million to 32.0 million over the window (20.8 percent). AI-exposed sectors grew at 2.15 percent per year against 3.42 percent for less-exposed sectors, a gap of 1.27 percentage points, and their share of the total fell from 18.3 to 17.2 percent. The sector-level difference in annualized growth was not significant (mean difference −0.63 percentage points, 95% CI −2.49 to 1.23). Conclusions: The one-person company grew fastest outside AI-exposed knowledge work, driven by transportation and platform-based services, which challenges the claim that generative AI is disproportionately enabling solo entrepreneurship. 3 03 06 10.62486/978-9915-9966-3-9.Ch03 01 04 3 Enterprise artificial intelligence adoption and solo self-employment in knowledge-intensive activities: an ecological study across European countries 1 A01 21 0009-0008-6878-0441 Rómulo Esteban Montilla Rómulo Esteban Montilla 01 eng 30 00 Introduction: Whether the diffusion of artificial intelligence among firms is associated with the level of one-person self-employment in knowledge-intensive activities is not established at the country level. Methods: An ecological cross-sectional study of the 27 European Union member states combined the 2024 share of enterprises using artificial intelligence (Eurostat) with the 2023 share of total employment accounted for by self-employed persons without employees in knowledge-intensive services, and estimated crude and GDP-adjusted associations. Results: Enterprise artificial intelligence adoption averaged 14.3 percent (SD 7.2) and solo self-employment in knowledge-intensive services averaged 2.0 percent of employment (SD 1.0). In the crude analysis the association was weak and not significant (β = 0.017 percentage points per percentage point of adoption, 95% CI −0.037 to 0.072, p = 0.524), and it attenuated after adjustment for gross domestic product per capita (β = −0.005, 95% CI −0.088 to 0.078, p = 0.898). Conclusions: Enterprise artificial intelligence adoption was not significantly associated with solo self-employment in knowledge-intensive activities across European countries, and the association was not confounded in a direction that would suggest a positive relationship after adjustment. 4 03 06 10.62486/978-9915-9966-3-9.Ch04 01 04 4 Individual and organizational creators of the most-downloaded open artificial intelligence models 1 A01 21 0000-0001-6029-8468 Dennyse María Patricia Hermosa Guzmán Dennyse María Patricia Hermosa Guzmán 01 eng 30 00 Introduction: Whether the most-used open artificial intelligence models are produced by organizations or by individual creators is an open question with direct consequences for funding and platform policy. Methods: The 500 most-downloaded public models on the Hugging Face Hub were retrieved and each author account was classified as an individual or an organization using the platform application programming interface. Results: Organizations authored 430 of the 500 models (86.0 percent) and 91.3 percent of their downloads, against 70 models (14.0 percent) and 8.7 percent of downloads by individuals. Ten individual creators reached the top, concentrated in speech recognition and embedding models, and their median downloads (6.55 million) matched the organizational median (6.47 million). Conclusions: The most-downloaded open models are organizationally dominated, but a small set of individual super-creators still reaches the top in specialized niches, which suggests targeted support for individuals is feasible rather than futile. 5 03 06 10.62486/978-9915-9966-3-9.Ch05 01 04 5 Single-authored papers in artificial intelligence: a time-trend analysis of arXiv metadata 1 A01 21 0000-0002-4300-9174 Daniel Roman-Acosta Daniel Roman-Acosta 01 eng 30 00 Introduction: Whether the solo researcher is being displaced by the growth of team science is a question that the artificial-intelligence literature has not answered with a time-trend estimate. Methods: The arXiv application programming interface was queried to draw a fixed sample of 200 papers per year from the cs.AI category for each year from 2018 to 2025, and the single-authored share was computed and regressed on year. Results: The single-authored share in cs.AI fell from 18.0 percent in 2018 to 11.0 percent in 2025, a trend of −0.536 percentage points per year (95% CI −1.910 to 0.838, p = 0.377). Under the cs.LG specification the trend was −0.399 percentage points per year and kept its sign. Conclusions: The share of single-authored artificial-intelligence papers declined over the window in all specifications, but the trend was not statistically significant, so the evidence is consistent with a slow erosion of the solo author rather than an abrupt displacement. 6 03 06 10.62486/978-9915-9966-3-9.Ch06 01 04 6 Contributor concentration in popular open-source artificial-intelligence agent frameworks 1 A01 21 0000-0003-0425-7201 Carlos Alberto Gómez Cano Carlos Alberto Gómez Cano 01 eng 30 00 Introduction: Open-source artificial-intelligence agent frameworks are infrastructure on which downstream applications depend, and their sustainability turns on whether they are maintained by one person or by a community. Methods: The twenty most-starred public repositories matching the query "AI agent framework" were retrieved from the GitHub application programming interface, and the share of commits contributed by the top contributor and by the top three contributors was computed for each. Results: The top contributor accounted for a median of 37.7 percent of commits, with a range from 11.7 to 91.2 percent, and five of the twenty repositories had a top contributor above 50 percent. The top three contributors accounted for a median of 57.6 percent of commits. Conclusions: Contribution to the most popular open-source agent frameworks is highly concentrated, and a substantial minority of these frameworks depend on a single dominant contributor, a maintenance risk that the field's reliance on them does not currently price in. 7 03 06 10.62486/978-9915-9966-3-9.Ch07 01 04 7 Generative artificial intelligence and individual worker productivity: a systematic review of experimental evidence 1 A01 21 0000-0001-6371-448X German Martínez Prats German Martínez Prats 01 eng 30 00 Introduction: The experimental literature on generative artificial intelligence and worker productivity has grown rapidly, but its results have not been synthesized by productivity dimension. Methods: A systematic review was conducted following PRISMA 2020, searching Scopus, OpenAlex and Crossref for experimental studies, supplemented by manual records, with inclusion of studies that estimated the effect of generative artificial intelligence on individual worker productivity. Results: Fourteen experimental studies were included. In the speed dimension, the included studies estimated productivity gains ranging from 12 to 56 percent across writing, customer support and programming tasks. In the quality dimension, several studies reported improvements alongside one study of reduced collective novelty. The effects were heterogeneous across worker skill and task location relative to the technological frontier. Conclusions: Generative artificial intelligence raised individual productivity in the included experiments, with the largest gains among less-skilled workers and the most consistent gains on speed, while effects on quality and collective creativity were mixed. 8 03 06 10.62486/978-9915-9966-3-9.Ch08 01 04 8 Determinants of artificial intelligence adoption in micro and small enterprises: a systematic review 1 A01 21 0000-0003-1935-2594 Yoan Hernández Flores Yoan Hernández Flores 01 eng 30 00 Introduction: Artificial intelligence adoption in micro and small enterprises is slower than in large firms, and the determinants of that gap have been studied piecemeal rather than synthesized. Methods: A systematic review was conducted following PRISMA 2020, were included studies of artificial intelligence adoption determinants in micro and small enterprises, with a narrative synthesis organized by the technological, organizational and environmental framework. Results: Thirteen studies were included. Organizational determinants, particularly top management support and digital leadership, and technological determinants, particularly perceived usefulness and compatibility, were the most consistently reported, while environmental determinants such as competitive and government pressure were reported less consistently. Conclusions: Artificial intelligence adoption in micro and small enterprises is driven most strongly by organizational and technological determinants, with environmental determinants secondary, which implies that interventions should target managerial capability and perceived usefulness rather than external incentives alone. 9 03 06 10.62486/978-9915-9966-3-9.Ch09 01 04 9 Generative artificial intelligence in entrepreneurship: a systematic review of empirical studies on opportunity recognition and venture creation 1 A01 21 0000-0002-3669-3123 Verenice Sánchez Castillo Verenice Sánchez Castillo 01 eng 30 00 Generative artificial intelligence is reshaping entrepreneurship, but the empirical evidence on its role in opportunity recognition and venture creation has not been synthesized. A systematic review was conducted following PRISMA methodology, were included empirical studies of generative artificial intelligence in entrepreneurship, with a narrative synthesis organized by opportunity recognition and venture creation. Eleven studies were included. The pattern split the evidence, with studies of opportunity recognition and decision-making reporting augmentation and, in one field experiment, a reduction in innovation, while studies of venture creation reported faster formation and stronger competitive positioning, with effects moderated by founder skill and task location. Generative artificial intelligence both enables and complicates the entrepreneurial process, augmenting opportunity recognition and venture creation for capable founders while introducing a risk of homogenization and judgment loss, so its net effect is conditional on the founder rather than uniformly positive. 10 03 06 10.62486/978-9915-9966-3-9.Ch10 01 04 10 From superstar firms to super-individuals: theoretical foundations of the one-person company 1 A01 21 0000-0002-5514-2992 Matilde Bolaño-García Matilde Bolaño-García 01 eng 30 00 The one-person company, a firm in which artificial intelligence substitutes for the employees, managers and contractors that a conventional venture would hire, is presented in policy and practitioner commentary as the successor to the superstar firm, but its theoretical foundations have not been assembled. This narrative review synthesizes the economics of superstar firms, the transaction-cost and knowledge-based theories of the firm, and the task-based automation literature to state the conditions under which the transition from the superstar firm to the super-individual is coherent and the conditions under which it fails. The synthesis shows that the superstar-firm literature explains why production concentrated, that the theory-of-the-firm literature explains which costs artificial intelligence lowers, and that the task-based literature explains why the founder, rather than the firm, becomes the unit of production. The one-person company hypothesis is coherent as an extension of these three bodies of work, but it is not yet supported by any verified case of a one-person unicorn, and its boundary conditions are founder skill, platform dependence and the irreplaceable residue of human judgment. The review concludes that the super-individual is a plausible limiting case of a real technological trend rather than an established fact, and that future work should test it against survival, scale and liability outcomes. 11 03 06 10.62486/978-9915-9966-3-9.Ch11 01 04 11 Multi-agent artificial intelligence systems as the workforce of the one-person company 1 A01 21 0000-0003-4555-7518 Alfredo Javier Pérez Gamboa Alfredo Javier Pérez Gamboa 01 eng 30 00 The one-person company, a firm in which a single founder is supported by artificial intelligence rather than by employees, depends on a specific technical claim, that a collection of autonomous agents can perform the work of a workforce. This narrative review synthesizes the architectures, the business functions and the failure modes of multi-agent artificial intelligence systems to assess that claim. The synthesis finds that multi-agent systems have moved from single large language models to orchestrated collections of specialized agents, that these collections now perform bounded business functions such as process automation, procurement, marketing and workflow orchestration, and that their failure modes, error propagation, collusion, security gaps and a missing oversight regime, bound the reliance that a founder can place on them. The review concludes that the multi-agent workforce is a real and rapidly maturing technical substrate, that its business functions are demonstrated for bounded workflows rather than for open-ended judgment, and that the solopreneur who adopts it inherits a set of oversight obligations that the technology does not discharge. 12 03 06 10.62486/978-9915-9966-3-9.Ch12 01 04 12 DOI: 10.62486/978-9915-9966-3-9.Ch12 Legal personality, liability and taxation of artificial-intelligence-run one-person companies 1 A01 21 0000-0003-2581-1873 Rolando Eslava Zapata Rolando Eslava Zapata 01 eng 30 00 The one-person company managed by artificial intelligence sits in the unresolved gap of the law of autonomous systems, because the law has not decided whether the system, the company or the founder is the legal person. This narrative review synthesizes the doctrinal debate on legal personality, the comparative frameworks of liability and the proposals for taxing autonomous systems to state where the law resolves the one-person company and where it does not. The synthesis finds that the electronic person remains contested and largely rejected in practice, that corporate personhood is the existing vehicle through which an artificial-intelligence-run company holds rights and duties, and that liability and taxation still anchor on a residual human who is, in the one-person company, a single founder. The review concludes that the one-person company is legally possible through the corporate form, that its liability and tax treatment remain anchored to the founder rather than to the machine, and that the unresolved question is not whether the machine can act but who answers for it. 13 03 06 10.62486/978-9915-9966-3-9.Ch13 01 04 13 The one-person unicorn hypothesis: systemic risks, platform dependence and future scenarios 1 A01 21 0000-0002-3681-9152 Juan Ernesto Pérez-Pérez Juan Ernesto Pérez-Pérez 01 eng 30 00 The one-person unicorn, a company that reaches unicorn scale with a single founder supported by artificial intelligence, is presented as a plausible near-term outcome, but its systemic risks, its dependence on platforms and its future scenarios have not been examined together. This narrative review synthesizes the evidence on the systemic risks, the platform dependence and the future scenarios of the one-person company to separate what is established from what is speculative. The synthesis finds that the systemic risk of the form is concentrated in the single founder, that the platform dependence is the mechanism through which the one-person company surrenders the value it captures, and that the future scenarios range from a dispersed super-individual to a re-concentrated platform economy, with no verified one-person unicorn to settle the range. The review concludes that the one-person unicorn is a coherent limit of real trends rather than an observed fact, that its platform dependence is the condition most likely to decide the hypothesis, and that the scenarios remain open until a verified case closes them. South American Publishing 01 South American Publishing 02 https://southam.pub/book-catalog/isbn/978-9915-9966-3-9.php Montevideo UY 04 01 20261030 01 WORLD 06 15 9789915996622 9789915996622-en-2 03 01 15 9789915996622 06 10.62486/978-9915-9966-3-9 00 BC 01 01 Superintelligence Series Volume 4: One Person Company: Artificial Intelligence, Autonomous Entrepreneurship and the Future of Business 1 B01 21 0000-0003-0425-7201 Carlos Alberto Gómez Cano Carlos Alberto Gómez Cano 1 01 eng 10 BUS025000 BUSINESS & ECONOMICS / Entrepreneurship 10 BUS115000 BUSINESS & ECONOMICS / Freelance & Self-Employment 10 COM100000 COMPUTERS / Artificial Intelligence / Generative AI 10 BUS041000 BUSINESS & ECONOMICS / Management 93 UYQE 20 Artificial intelligence; One-person companies; Entrepreneurship; Solo self-employment; Multi-agent systems; Platform dependence 30 00 The possibility of an enterprise operated by a single founder with extensive support from artificial intelligence raises questions about business formation, the organization of work, and the distribution of economic responsibility. Bringing together thirteen chapters, this volume examines the empirical basis and theoretical limits of the one-person company. Studies of business applications, nonemployer establishments, and European self-employment patterns find no consistent evidence of a broad entrepreneurial surge attributable to AI diffusion. Analyses of open models, research authorship, and software repositories show that individual contributions remain possible, although prominent technical resources often depend on organizations or concentrated groups of maintainers. Reviews of experimental productivity research identify gains in the speed of selected tasks, with more variable effects on quality and creativity. Research on smaller enterprises and venture creation points to managerial capability, founder judgment, and the fit between tools and tasks as conditions shaping adoption and outcomes. The volume also considers how theories of the firm can account for AI-assisted production, what multi-agent systems can accomplish in bounded workflows, and why legal personality, liability, taxation, and platform dependence remain unresolved or consequential. The prospect of a one-person unicorn is treated as an unverified hypothesis rather than an observed business model. Across these perspectives, the central distinction is between automating activities and sustaining an accountable enterprise. Artificial intelligence may expand what an individual can accomplish, but the viability and independence of a firm continue to depend on institutional arrangements, oversight, and the distribution of risk. 04 00 Chapter 1: Business formation and the public release of generative artificial intelligence: an interrupted time-series analysis; Chapter 2: Nonemployer business growth in artificial-intelligence-exposed industries of the United States; Chapter 3: Enterprise artificial intelligence adoption and solo self-employment in knowledge-intensive activities: an ecological study across European countries; Chapter 4: Individual and organizational creators of the most-downloaded open artificial intelligence models; Chapter 5: Single-authored papers in artificial intelligence: a time-trend analysis of arXiv metadata; Chapter 6: Contributor concentration in popular open-source artificial-intelligence agent frameworks; Chapter 7: Generative artificial intelligence and individual worker productivity: a systematic review of experimental evidence; Chapter 8: Determinants of artificial intelligence adoption in micro and small enterprises: a systematic review; Chapter 9: Generative artificial intelligence in entrepreneurship: a systematic review of empirical studies on opportunity recognition and venture creation; Chapter 10: From superstar firms to super-individuals: theoretical foundations of the one-person company; Chapter 11: Multi-agent artificial intelligence systems as the workforce of the one-person company; Chapter 12: DOI: 10.62486/978-9915-9966-3-9.Ch12 Legal personality, liability and taxation of artificial-intelligence-run one-person companies; Chapter 13: The one-person unicorn hypothesis: systemic risks, platform dependence and future scenarios 01 00 03 02 https://southam.pub/book-catalog/2026/978-9915-9966-3-9/978-9915-9966-3-9_Portada.png 1 03 06 10.62486/978-9915-9966-3-9.Ch01 01 04 1 Business formation and the public release of generative artificial intelligence: an interrupted time-series analysis 1 A01 21 0000-0002-9455-4324 Libardo Ramón Polanía Libardo Ramón Polanía 2 A01 21 0009-0008-6878-0441 Rómulo Esteban Montilla Rómulo Esteban Montilla 01 eng 30 00 Introduction: The public release of generative artificial intelligence in November 2022 promised to lower the cost of starting and running a firm, yet its effect on aggregate business formation remains unmeasured. Methods: We used monthly business applications from the US Census Bureau Business Formation Statistics, seasonally adjusted, from January 2018 through August 2026, and fitted an interrupted time-series model with a break at November 2022, estimating level and trend changes with Newey-West standard errors (12 lags) for the aggregate and for three knowledge-intensive sectors. Results: At the national level the immediate shift was −49,832 applications (p = 0.099) and the monthly trend slowed by 1,602 applications (p = 0.034), from a pre-release slope of 3,631. Sectoral responses diverged, with Information falling 2,123 applications at the break (p = 0.032) and Management of companies accelerating by 44 applications per month (p < 0.001). Conclusions: The release of generative artificial intelligence coincided with a slowdown in aggregate business formation and with divergent, sector-specific responses, a pattern consistent with skill-biased reallocation rather than a uniform startup surge. 2 03 06 10.62486/978-9915-9966-3-9.Ch02 01 04 2 Nonemployer business growth in artificial-intelligence-exposed industries of the United States 1 A01 21 0009-0002-0970-2810 Edixon Chacón Guerrero Edixon Chacón Guerrero 01 eng 30 00 Introduction: Whether generative artificial intelligence enables the one-person company or instead concentrates advantage in larger employers is an open empirical question. Methods: We used national nonemployer establishment counts from the US Census Bureau Nonemployer Statistics, 2018 to 2024, aggregated to two-digit NAICS sectors, and compared compound annual growth between AI-exposed sectors (Information, Finance and insurance, Professional services) and the remaining sectors. Results: Nonemployer establishments rose from 26.5 million to 32.0 million over the window (20.8 percent). AI-exposed sectors grew at 2.15 percent per year against 3.42 percent for less-exposed sectors, a gap of 1.27 percentage points, and their share of the total fell from 18.3 to 17.2 percent. The sector-level difference in annualized growth was not significant (mean difference −0.63 percentage points, 95% CI −2.49 to 1.23). Conclusions: The one-person company grew fastest outside AI-exposed knowledge work, driven by transportation and platform-based services, which challenges the claim that generative AI is disproportionately enabling solo entrepreneurship. 3 03 06 10.62486/978-9915-9966-3-9.Ch03 01 04 3 Enterprise artificial intelligence adoption and solo self-employment in knowledge-intensive activities: an ecological study across European countries 1 A01 21 0009-0008-6878-0441 Rómulo Esteban Montilla Rómulo Esteban Montilla 01 eng 30 00 Introduction: Whether the diffusion of artificial intelligence among firms is associated with the level of one-person self-employment in knowledge-intensive activities is not established at the country level. Methods: An ecological cross-sectional study of the 27 European Union member states combined the 2024 share of enterprises using artificial intelligence (Eurostat) with the 2023 share of total employment accounted for by self-employed persons without employees in knowledge-intensive services, and estimated crude and GDP-adjusted associations. Results: Enterprise artificial intelligence adoption averaged 14.3 percent (SD 7.2) and solo self-employment in knowledge-intensive services averaged 2.0 percent of employment (SD 1.0). In the crude analysis the association was weak and not significant (β = 0.017 percentage points per percentage point of adoption, 95% CI −0.037 to 0.072, p = 0.524), and it attenuated after adjustment for gross domestic product per capita (β = −0.005, 95% CI −0.088 to 0.078, p = 0.898). Conclusions: Enterprise artificial intelligence adoption was not significantly associated with solo self-employment in knowledge-intensive activities across European countries, and the association was not confounded in a direction that would suggest a positive relationship after adjustment. 4 03 06 10.62486/978-9915-9966-3-9.Ch04 01 04 4 Individual and organizational creators of the most-downloaded open artificial intelligence models 1 A01 21 0000-0001-6029-8468 Dennyse María Patricia Hermosa Guzmán Dennyse María Patricia Hermosa Guzmán 01 eng 30 00 Introduction: Whether the most-used open artificial intelligence models are produced by organizations or by individual creators is an open question with direct consequences for funding and platform policy. Methods: The 500 most-downloaded public models on the Hugging Face Hub were retrieved and each author account was classified as an individual or an organization using the platform application programming interface. Results: Organizations authored 430 of the 500 models (86.0 percent) and 91.3 percent of their downloads, against 70 models (14.0 percent) and 8.7 percent of downloads by individuals. Ten individual creators reached the top, concentrated in speech recognition and embedding models, and their median downloads (6.55 million) matched the organizational median (6.47 million). Conclusions: The most-downloaded open models are organizationally dominated, but a small set of individual super-creators still reaches the top in specialized niches, which suggests targeted support for individuals is feasible rather than futile. 5 03 06 10.62486/978-9915-9966-3-9.Ch05 01 04 5 Single-authored papers in artificial intelligence: a time-trend analysis of arXiv metadata 1 A01 21 0000-0002-4300-9174 Daniel Roman-Acosta Daniel Roman-Acosta 01 eng 30 00 Introduction: Whether the solo researcher is being displaced by the growth of team science is a question that the artificial-intelligence literature has not answered with a time-trend estimate. Methods: The arXiv application programming interface was queried to draw a fixed sample of 200 papers per year from the cs.AI category for each year from 2018 to 2025, and the single-authored share was computed and regressed on year. Results: The single-authored share in cs.AI fell from 18.0 percent in 2018 to 11.0 percent in 2025, a trend of −0.536 percentage points per year (95% CI −1.910 to 0.838, p = 0.377). Under the cs.LG specification the trend was −0.399 percentage points per year and kept its sign. Conclusions: The share of single-authored artificial-intelligence papers declined over the window in all specifications, but the trend was not statistically significant, so the evidence is consistent with a slow erosion of the solo author rather than an abrupt displacement. 6 03 06 10.62486/978-9915-9966-3-9.Ch06 01 04 6 Contributor concentration in popular open-source artificial-intelligence agent frameworks 1 A01 21 0000-0003-0425-7201 Carlos Alberto Gómez Cano Carlos Alberto Gómez Cano 01 eng 30 00 Introduction: Open-source artificial-intelligence agent frameworks are infrastructure on which downstream applications depend, and their sustainability turns on whether they are maintained by one person or by a community. Methods: The twenty most-starred public repositories matching the query "AI agent framework" were retrieved from the GitHub application programming interface, and the share of commits contributed by the top contributor and by the top three contributors was computed for each. Results: The top contributor accounted for a median of 37.7 percent of commits, with a range from 11.7 to 91.2 percent, and five of the twenty repositories had a top contributor above 50 percent. The top three contributors accounted for a median of 57.6 percent of commits. Conclusions: Contribution to the most popular open-source agent frameworks is highly concentrated, and a substantial minority of these frameworks depend on a single dominant contributor, a maintenance risk that the field's reliance on them does not currently price in. 7 03 06 10.62486/978-9915-9966-3-9.Ch07 01 04 7 Generative artificial intelligence and individual worker productivity: a systematic review of experimental evidence 1 A01 21 0000-0001-6371-448X German Martínez Prats German Martínez Prats 01 eng 30 00 Introduction: The experimental literature on generative artificial intelligence and worker productivity has grown rapidly, but its results have not been synthesized by productivity dimension. Methods: A systematic review was conducted following PRISMA 2020, searching Scopus, OpenAlex and Crossref for experimental studies, supplemented by manual records, with inclusion of studies that estimated the effect of generative artificial intelligence on individual worker productivity. Results: Fourteen experimental studies were included. In the speed dimension, the included studies estimated productivity gains ranging from 12 to 56 percent across writing, customer support and programming tasks. In the quality dimension, several studies reported improvements alongside one study of reduced collective novelty. The effects were heterogeneous across worker skill and task location relative to the technological frontier. Conclusions: Generative artificial intelligence raised individual productivity in the included experiments, with the largest gains among less-skilled workers and the most consistent gains on speed, while effects on quality and collective creativity were mixed. 8 03 06 10.62486/978-9915-9966-3-9.Ch08 01 04 8 Determinants of artificial intelligence adoption in micro and small enterprises: a systematic review 1 A01 21 0000-0003-1935-2594 Yoan Hernández Flores Yoan Hernández Flores 01 eng 30 00 Introduction: Artificial intelligence adoption in micro and small enterprises is slower than in large firms, and the determinants of that gap have been studied piecemeal rather than synthesized. Methods: A systematic review was conducted following PRISMA 2020, were included studies of artificial intelligence adoption determinants in micro and small enterprises, with a narrative synthesis organized by the technological, organizational and environmental framework. Results: Thirteen studies were included. Organizational determinants, particularly top management support and digital leadership, and technological determinants, particularly perceived usefulness and compatibility, were the most consistently reported, while environmental determinants such as competitive and government pressure were reported less consistently. Conclusions: Artificial intelligence adoption in micro and small enterprises is driven most strongly by organizational and technological determinants, with environmental determinants secondary, which implies that interventions should target managerial capability and perceived usefulness rather than external incentives alone. 9 03 06 10.62486/978-9915-9966-3-9.Ch09 01 04 9 Generative artificial intelligence in entrepreneurship: a systematic review of empirical studies on opportunity recognition and venture creation 1 A01 21 0000-0002-3669-3123 Verenice Sánchez Castillo Verenice Sánchez Castillo 01 eng 30 00 Generative artificial intelligence is reshaping entrepreneurship, but the empirical evidence on its role in opportunity recognition and venture creation has not been synthesized. A systematic review was conducted following PRISMA methodology, were included empirical studies of generative artificial intelligence in entrepreneurship, with a narrative synthesis organized by opportunity recognition and venture creation. Eleven studies were included. The pattern split the evidence, with studies of opportunity recognition and decision-making reporting augmentation and, in one field experiment, a reduction in innovation, while studies of venture creation reported faster formation and stronger competitive positioning, with effects moderated by founder skill and task location. Generative artificial intelligence both enables and complicates the entrepreneurial process, augmenting opportunity recognition and venture creation for capable founders while introducing a risk of homogenization and judgment loss, so its net effect is conditional on the founder rather than uniformly positive. 10 03 06 10.62486/978-9915-9966-3-9.Ch10 01 04 10 From superstar firms to super-individuals: theoretical foundations of the one-person company 1 A01 21 0000-0002-5514-2992 Matilde Bolaño-García Matilde Bolaño-García 01 eng 30 00 The one-person company, a firm in which artificial intelligence substitutes for the employees, managers and contractors that a conventional venture would hire, is presented in policy and practitioner commentary as the successor to the superstar firm, but its theoretical foundations have not been assembled. This narrative review synthesizes the economics of superstar firms, the transaction-cost and knowledge-based theories of the firm, and the task-based automation literature to state the conditions under which the transition from the superstar firm to the super-individual is coherent and the conditions under which it fails. The synthesis shows that the superstar-firm literature explains why production concentrated, that the theory-of-the-firm literature explains which costs artificial intelligence lowers, and that the task-based literature explains why the founder, rather than the firm, becomes the unit of production. The one-person company hypothesis is coherent as an extension of these three bodies of work, but it is not yet supported by any verified case of a one-person unicorn, and its boundary conditions are founder skill, platform dependence and the irreplaceable residue of human judgment. The review concludes that the super-individual is a plausible limiting case of a real technological trend rather than an established fact, and that future work should test it against survival, scale and liability outcomes. 11 03 06 10.62486/978-9915-9966-3-9.Ch11 01 04 11 Multi-agent artificial intelligence systems as the workforce of the one-person company 1 A01 21 0000-0003-4555-7518 Alfredo Javier Pérez Gamboa Alfredo Javier Pérez Gamboa 01 eng 30 00 The one-person company, a firm in which a single founder is supported by artificial intelligence rather than by employees, depends on a specific technical claim, that a collection of autonomous agents can perform the work of a workforce. This narrative review synthesizes the architectures, the business functions and the failure modes of multi-agent artificial intelligence systems to assess that claim. The synthesis finds that multi-agent systems have moved from single large language models to orchestrated collections of specialized agents, that these collections now perform bounded business functions such as process automation, procurement, marketing and workflow orchestration, and that their failure modes, error propagation, collusion, security gaps and a missing oversight regime, bound the reliance that a founder can place on them. The review concludes that the multi-agent workforce is a real and rapidly maturing technical substrate, that its business functions are demonstrated for bounded workflows rather than for open-ended judgment, and that the solopreneur who adopts it inherits a set of oversight obligations that the technology does not discharge. 12 03 06 10.62486/978-9915-9966-3-9.Ch12 01 04 12 DOI: 10.62486/978-9915-9966-3-9.Ch12 Legal personality, liability and taxation of artificial-intelligence-run one-person companies 1 A01 21 0000-0003-2581-1873 Rolando Eslava Zapata Rolando Eslava Zapata 01 eng 30 00 The one-person company managed by artificial intelligence sits in the unresolved gap of the law of autonomous systems, because the law has not decided whether the system, the company or the founder is the legal person. This narrative review synthesizes the doctrinal debate on legal personality, the comparative frameworks of liability and the proposals for taxing autonomous systems to state where the law resolves the one-person company and where it does not. The synthesis finds that the electronic person remains contested and largely rejected in practice, that corporate personhood is the existing vehicle through which an artificial-intelligence-run company holds rights and duties, and that liability and taxation still anchor on a residual human who is, in the one-person company, a single founder. The review concludes that the one-person company is legally possible through the corporate form, that its liability and tax treatment remain anchored to the founder rather than to the machine, and that the unresolved question is not whether the machine can act but who answers for it. 13 03 06 10.62486/978-9915-9966-3-9.Ch13 01 04 13 The one-person unicorn hypothesis: systemic risks, platform dependence and future scenarios 1 A01 21 0000-0002-3681-9152 Juan Ernesto Pérez-Pérez Juan Ernesto Pérez-Pérez 01 eng 30 00 The one-person unicorn, a company that reaches unicorn scale with a single founder supported by artificial intelligence, is presented as a plausible near-term outcome, but its systemic risks, its dependence on platforms and its future scenarios have not been examined together. This narrative review synthesizes the evidence on the systemic risks, the platform dependence and the future scenarios of the one-person company to separate what is established from what is speculative. The synthesis finds that the systemic risk of the form is concentrated in the single founder, that the platform dependence is the mechanism through which the one-person company surrenders the value it captures, and that the future scenarios range from a dispersed super-individual to a re-concentrated platform economy, with no verified one-person unicorn to settle the range. The review concludes that the one-person unicorn is a coherent limit of real trends rather than an observed fact, that its platform dependence is the condition most likely to decide the hypothesis, and that the scenarios remain open until a verified case closes them. South American Publishing 01 South American Publishing 02 https://southam.pub/book-catalog/isbn/978-9915-9966-3-9.php Montevideo UY 04 01 20261030 01 WORLD 06 15 9789915996639