|
|
| Line 3: |
Line 3: |
| Artificial intelligence depends upon extensive physical and social infrastructure, including [[semiconductor]] production, data centers, electrical grids, telecommunications networks, software, datasets, mineral extraction, and human labor. Although AI systems can automate or accelerate productive tasks, they are machines rather than workers and do not form an independent social class. | | Artificial intelligence depends upon extensive physical and social infrastructure, including [[semiconductor]] production, data centers, electrical grids, telecommunications networks, software, datasets, mineral extraction, and human labor. Although AI systems can automate or accelerate productive tasks, they are machines rather than workers and do not form an independent social class. |
|
| |
|
| To [[Marxism|Marxists]], artificial intelligence may be understood as part of the [[means of production]]. Under [[capitalism]], AI generally functions as privately owned [[constant capital]] and may be used to increase productivity, reduce labor costs, monitor workers, collect data, expand monopoly control, and strengthen the power of capital over the labor process. AI and robots do not sell their [[labor power]] and do not independently create [[value]] in the Marxist sense, although their development and operation depend upon large amounts of living labor.
| | Artificial intelligence is a [[means of production]]. Under [[capitalism]], AI is privately owned [[constant capital]]; AI, robots and machines do not sell their [[labor power]] and do not independently create [[value]], although their development and operation depend upon large amounts of [[living labor]]. |
|
| |
|
| The social effects of AI are not determined by the technology alone. They also depend upon ownership, political authority, labor relations, energy systems, regulation, and the purposes toward which automated systems are directed. AI may be used for military targeting, policing, workplace surveillance, political propaganda, and commercial exploitation, but it may also support scientific research, accessibility, translation, education, healthcare, communication, creative production, and the reduction of dangerous or repetitive labor. | | The social effects of AI are not determined by the technology alone, but on the broader question of which [[Social class|class]] owns constant capital (at large in society). While AI may be used for military purpose, policing, surveillance and other anti-communist purposes, it may also support scientific research, accessibility, translation, education, healthcare, communication, creative production, and the reduction of dangerous or repetitive labor. |
| | |
| [[Techno-socialism|Techno-socialists]] argue that artificial intelligence and digital infrastructure should be placed under public, cooperative, or worker control so that increased productivity produces shorter working hours, greater accessibility, improved public services, and broadly distributed social benefits rather than unemployment and private accumulation.
| |
|
| |
|
| == History == | | == History == |
| Line 15: |
Line 13: |
|
| |
|
| Generative AI systems became widely used during the 2020s. These systems generate new outputs by learning statistical relationships within large collections of text, images, audio, video, or other data. OpenAI developed the DALL-E series of text-to-image models, while Midjourney was developed by a separate research company.<ref>{{Web citation|title=DALL·E: Creating images from text|date=5 January 2021|url=https://openai.com/index/dall-e/|retrieved=1 August 2026|website=OpenAI}}</ref><ref>{{Web citation|title=Midjourney|url=https://www.midjourney.com/|retrieved=1 August 2026|website=Midjourney}}</ref> | | Generative AI systems became widely used during the 2020s. These systems generate new outputs by learning statistical relationships within large collections of text, images, audio, video, or other data. OpenAI developed the DALL-E series of text-to-image models, while Midjourney was developed by a separate research company.<ref>{{Web citation|title=DALL·E: Creating images from text|date=5 January 2021|url=https://openai.com/index/dall-e/|retrieved=1 August 2026|website=OpenAI}}</ref><ref>{{Web citation|title=Midjourney|url=https://www.midjourney.com/|retrieved=1 August 2026|website=Midjourney}}</ref> |
|
| |
| Large language models and image generators expanded public discussion over automation, creative labor, education, surveillance, military applications, environmental effects, and the concentration of technological infrastructure in a small number of corporations and states.
| |
|
| |
| == Political economy ==
| |
|
| |
| === AI as a means of production ===
| |
| The [[means of production]] are the tools, machinery, resources, and infrastructure through which society produces goods and services. In earlier industrial periods these included farms, mines, factories, railways, electrical systems, and heavy machinery. In the digital economy they also include computers, networks, algorithms, datasets, cloud platforms, data centers, robotics, and semiconductor infrastructure.
| |
|
| |
| AI increasingly forms part of productive activity because it can assist with writing, translation, software development, research, design, scheduling, analysis, communication, logistics, and administrative labor. It is therefore not merely a consumer novelty; in many industries it operates as productive machinery.
| |
|
| |
| Under capitalism, AI systems normally belong to corporations, investors, military agencies, or other institutions possessing sufficient capital to build or rent the infrastructure required to operate them. The political question is consequently not only whether AI should exist, but who owns its productive capacity, who determines its use, and who receives the gains created by greater productivity.
| |
|
| |
| === Constant capital and human labor ===
| |
| In Marxist political economy, machinery, buildings, raw materials, and infrastructure form [[constant capital]], while wages paid to workers purchase [[variable capital]], or labor power. AI can increase the productivity of labor, alter the organization of production, and reduce the amount of direct labor required for particular tasks.
| |
|
| |
| AI systems do not constitute an independent social class and do not receive wages. Their operation nevertheless depends upon workers involved in programming, data preparation, data labeling, content moderation, semiconductor fabrication, mineral extraction, electrical production, construction, maintenance, logistics, and communications. Generated outputs also draw upon cultural and scientific knowledge produced through previous human activity.
| |
|
| |
| As firms increase investment in machinery relative to living labor, Marx argued that competition and the rising organic composition of capital can create pressure upon the [[tendency of the rate of profit to fall]]. This tendency is not automatic or uninterrupted because capitalists employ counteracting measures, including intensified exploitation, cheaper inputs, expansion into new markets, and the destruction or devaluation of capital.<ref>{{Web citation|author=Karl Marx|title=Capital, Volume III, Part III: The Law of the Tendency of the Rate of Profit to Fall|url=https://www.marxists.org/archive/marx/works/1894-c3/ch13.htm|retrieved=1 August 2026|website=Marxists Internet Archive}}</ref>
| |
|
| |
| === Ownership and class power ===
| |
| Artificial intelligence is often described as though it independently chooses its social purpose. In practice, institutions determine which systems are funded, trained, deployed, and integrated into workplaces or public services.
| |
|
| |
| Under private ownership, AI may be used to reduce labor costs, increase surveillance, weaken worker bargaining power, collect personal information, privatize access to knowledge, and strengthen monopolies. Executives, owners, investors, and governments establish the labor and business policies surrounding automated systems.
| |
|
| |
| Socialist and labor-oriented proposals instead emphasize public ownership, cooperative control, union participation, democratic regulation, and the distribution of productivity gains among workers and society. From this perspective, the central contradiction is not between humanity and machinery, but between social production and private control.
| |
|
| |
|
| == Labor and employment == | | == Labor and employment == |
| | | The International Labour Organization reported in 2025 that one in four jobs worldwide had some exposure to generative AI, while concluding that transformation rather than complete replacement was the most likely overall outcome.<ref>{{Web citation|title=One in four jobs at risk of being transformed by GenAI, new ILO–NASK Global Index shows|date=20 May 2025|url=https://www.ilo.org/resource/news/one-four-jobs-risk-being-transformed-genai-new-ilo%E2%80%93nask-global-index-shows|retrieved=1 August 2026|website=International Labour Organization}}</ref> |
| === Automation of tasks and occupations ===
| |
| AI may automate entire occupations in some cases, but more commonly automates or accelerates particular tasks within a broader job. The International Labour Organization reported in 2025 that one in four jobs worldwide had some exposure to generative AI, while concluding that transformation rather than complete replacement was the most likely overall outcome.<ref>{{Web citation|title=One in four jobs at risk of being transformed by GenAI, new ILO–NASK Global Index shows|date=20 May 2025|url=https://www.ilo.org/resource/news/one-four-jobs-risk-being-transformed-genai-new-ilo%E2%80%93nask-global-index-shows|retrieved=1 August 2026|website=International Labour Organization}}</ref>
| |
| | |
| A job is usually a collection of tasks rather than a single indivisible activity. AI may automate drafting, classification, transcription, scheduling, or pattern recognition while leaving responsibility, physical work, interpersonal communication, contextual judgment, supervision, and decision-making to people.
| |
| | |
| Under capitalist ownership, automation may be used to dismiss workers, intensify labor, suppress wages, or weaken bargaining power. Labor and socialist proposals include shorter working hours without reduced pay, employment protections, publicly funded retraining, union participation in deployment decisions, and the social distribution of productivity gains.
| |
| | |
| === Deskilling and labor discipline ===
| |
| Automation can simplify portions of a job and transfer knowledge from workers into machinery or management systems. This may reduce workers' control over the labor process and allow employers to replace experienced employees with lower-paid labor. It may also increase managerial monitoring by measuring output, ranking employees, or assigning tasks through algorithms.
| |
| | |
| AI can alternatively assist workers by reducing repetitive labor, increasing access to information, or making specialized tools easier to use. Whether this produces greater autonomy or greater managerial control depends upon ownership, workplace organization, and worker power.
| |
| | |
| === Workplace surveillance ===
| |
| Employers use algorithmic systems to measure productivity, screen job applicants, monitor communications, track movement, assign schedules, and evaluate performance. Such systems can conceal managerial decisions behind technical language and make it difficult for workers to understand or challenge the basis of an employment decision.
| |
| | |
| A techno-socialist labor program would require transparency, collective bargaining, human review, appeal rights, and worker participation before automated systems are introduced into a workplace. It would also prohibit the use of AI to discipline or dismiss workers through unaccountable scoring systems.
| |
| | |
| === Automation and reduced labor time ===
| |
| Automation has the potential to reduce dangerous, exhausting, repetitive, and monotonous labor. The political question is what happens to the time saved by greater productivity.
| |
| | |
| Under capitalist ownership, productivity increases may result in layoffs, work intensification, precarious employment, and wealth concentration. Under a worker-centered system, the same gains could support shorter workweeks, equal or higher pay, more leisure, improved public services, and greater freedom.
| |
| | |
| From this perspective, if a machine saves labor time, the benefit should be returned to workers rather than appropriated entirely by owners. Automation should reduce necessary labor rather than human dignity.
| |
|
| |
|
| === WGA and SAG-AFTRA strikes === | | === WGA and SAG-AFTRA strikes === |
| Line 73: |
Line 23: |
|
| |
|
| SAG-AFTRA described its 2023 agreement as containing protections governing digital replicas and AI uses involving performers, including requirements concerning consent and compensation.<ref>{{Web citation|title=Artificial Intelligence|url=https://www.sagaftra.org/contracts-industry-resources/member-resources/artificial-intelligence|retrieved=1 August 2026|website=SAG-AFTRA}}</ref> | | SAG-AFTRA described its 2023 agreement as containing protections governing digital replicas and AI uses involving performers, including requirements concerning consent and compensation.<ref>{{Web citation|title=Artificial Intelligence|url=https://www.sagaftra.org/contracts-industry-resources/member-resources/artificial-intelligence|retrieved=1 August 2026|website=SAG-AFTRA}}</ref> |
|
| |
| The strikes demonstrated that opposition to specific corporate uses of AI does not necessarily require opposition to the technology in every context. The unions sought contractual control over deployment, consent, compensation, and working conditions.
| |
|
| |
| == Accessibility ==
| |
| Artificial intelligence is used in assistive systems including automated captioning, speech recognition, text-to-speech software, image description, translation, predictive communication, reading assistance, planning tools, and adaptive computer interfaces.
| |
|
| |
| These systems may support people with visual, auditory, cognitive, developmental, linguistic, learning, or motor disabilities. AI-assisted tools may help users organize information, communicate ideas, interpret complex language, access educational material, and participate in cultural production.
| |
|
| |
| Some disabled and neurodivergent users describe generative AI as a means of reducing barriers created by inaccessible institutions. Blanket opposition to AI that ignores such uses has consequently been criticized as ableist or socially exclusionary.
| |
|
| |
| AI accessibility tools may nevertheless contain errors, reproduce bias, collect sensitive information, or be used by institutions to replace properly funded human assistance. Advocates therefore emphasize user control, privacy, accessible design, and the participation of disabled people in technological development.
| |
|
| |
| === Education and communication ===
| |
| AI may provide tutoring, translation, brainstorming, organization, explanation, practice questions, feedback, and accessibility support. It may help students understand material, particularly where schools do not provide sufficient individual assistance.
| |
|
| |
| Submitting generated work as evidence of knowledge a student does not possess, fabricating citations, or replacing research with unsupported outputs is academically dishonest. The misuse lies in how a tool is employed rather than in the mere existence of the tool.
| |
|
| |
| Educational policies increasingly distinguish between assistance and substitution. Responsible use requires verification, disclosure where required, source evaluation, citation practices, privacy awareness, and an understanding that AI output is not itself reliable evidence.
| |
|
| |
| == Art and cultural production ==
| |
|
| |
| === Generative art ===
| |
| Generative AI systems can produce text, images, music, audio, video, and other media in response to human instructions. OpenAI's DALL-E and the separately developed Midjourney service became prominent image-generation systems during the 2020s.<ref>{{Web citation|title=DALL·E 3 system card|url=https://openai.com/index/dall-e-3-system-card/|retrieved=1 August 2026|website=OpenAI}}</ref><ref>{{Web citation|title=Getting Started Guide|url=https://docs.midjourney.com/hc/en-us/articles/33329261836941-Getting-Started-Guide|retrieved=1 August 2026|website=Midjourney Documentation}}</ref>
| |
|
| |
| AI-assisted art ranges from minimally directed generation to projects involving extensive prompting, revision, compositing, painting, editing, photography, design, and other human labor. The amount of human direction and technical work varies considerably.
| |
|
| |
| === Training data and intellectual property ===
| |
| Generative AI has produced disputes over authorship, consent, compensation, copyright, employment, attribution, and the use of creative works in training datasets.
| |
|
| |
| Supporters argue that generative systems participate in the cumulative and transformative character of culture and can expand access to creative production. Critics argue that many systems were built through the mass collection of works without meaningful consent and may be used to imitate creators, reduce employment, weaken bargaining power, or concentrate control within technology corporations.
| |
|
| |
| A socialist approach may distinguish between personal or assistive use, publicly governed cultural tools, and commercial systems based on uncompensated extraction. Proposals include union protections, transparency concerning datasets, collective licensing, limits on unauthorized digital replicas, public creative infrastructure, and restrictions on corporate monopoly ownership.
| |
|
| |
| === Human direction and AI-assisted production ===
| |
| AI does not operate as an independent creative consciousness. Human beings choose goals, provide instructions, select outputs, revise errors, combine materials, establish context, and determine how a finished work is presented.
| |
|
| |
| Prompting alone may require little labor, while other AI-assisted projects involve extensive direction and editing. These forms should not be treated as identical. Human authorship may involve conception, composition, interpretation, symbolism, selection, revision, and communication rather than the manual performance of every technical step.
| |
|
| |
| Photography, synthesizers, digital editing, word processors, and personal computers changed creative processes without eliminating all previous forms. AI is likely to coexist with manual, digital, and collaborative production, although its effects upon employment and cultural markets remain contested.
| |
|
| |
| === Creative labor and compensation ===
| |
| Artists, writers, actors, musicians, and other cultural workers have legitimate interests in compensation, credit, bargaining power, transparency, and protection against deceptive or unauthorized commercial uses.
| |
|
| |
| Anti-AI rhetoric is not automatically pro-worker. Corporations may invoke the defense of creativity while underpaying workers, controlling distribution, owning large cultural archives, or using copyright to preserve monopoly power. Conversely, support for personal AI use does not justify unrestricted corporate extraction or the replacement of unionized workers.
| |
|
| |
| == Environmental effects ==
| |
| Artificial intelligence depends upon data centers, semiconductor manufacturing, electrical generation, cooling systems, mineral extraction, telecommunications networks, and electronic hardware. Its environmental effects include electricity consumption, water use, emissions, construction impacts, and electronic waste.
| |
|
| |
| The scale of these effects varies according to the model, hardware, data center, electricity source, cooling method, number of users, and purpose of deployment. Training large models can require substantial computing resources, while the cumulative cost of repeated inference may also become significant.
| |
|
| |
| The International Energy Agency reported that there is no AI without electricity for data centers and projected substantial growth in data-center energy demand as AI deployment expands.<ref>{{Web citation|title=Energy and AI|date=April 2025|url=https://www.iea.org/reports/energy-and-ai/|retrieved=1 August 2026|website=International Energy Agency}}</ref>
| |
|
| |
| === Energy and public planning ===
| |
| Energy consumption and energy source are distinct questions. A computing system powered by coal or gas has a different emissions profile from one supplied by lower-carbon electricity. Public planning, efficiency standards, improved cooling, shared infrastructure, and limits upon wasteful deployment can reduce environmental harm.
| |
|
| |
| A socialist environmental analysis examines corporate investment, grid planning, industrial supply chains, planned obsolescence, and the social purposes for which computing resources are allocated. It does not reduce responsibility to individual users making prompts.
| |
|
| |
| === Criticism of consumer-focused environmentalism ===
| |
| Criticism of AI's environmental effects becomes selective when individual prompts are treated as uniquely destructive while fossil-fuel extraction, military production, luxury consumption, advertising, cryptocurrency speculation, planned obsolescence, and the wider cloud economy receive less attention.
| |
|
| |
| This does not make AI environmentally harmless. It means ecological analysis should follow ownership, energy infrastructure, corporate expansion, and industrial policy rather than relying primarily upon consumer guilt.
| |
|
| |
| == Surveillance and policing ==
| |
| Police and security agencies use AI for facial recognition, automated license-plate recognition, predictive systems, social-media analysis, pattern detection, and the classification of people or locations as risks.
| |
|
| |
| Automated systems can reproduce racial and class inequalities contained in policing data. When historical arrest patterns are treated as neutral evidence of criminality, algorithms may direct additional police attention toward communities already subjected to disproportionate surveillance.
| |
|
| |
| AI can intensify surveillance by processing information more quickly and connecting previously separate databases. The underlying problem is the concentration of data and coercive authority within police agencies, governments, landlords, employers, and corporations.
| |
|
| |
| Techno-socialist proposals include restrictions or prohibitions on predictive policing, bans on abusive biometric surveillance, limits on data brokerage, transparent public oversight, and rights to explanation and appeal.
| |
|
| |
|
| == Military applications == | | == Military applications == |
| Military institutions use or research AI for surveillance, target recognition, logistics, autonomous navigation, intelligence analysis, cyber operations, propaganda, and weapons systems.
| |
|
| |
| In 2023, a United States Air Force officer described a hypothetical scenario in which an AI-controlled drone attacked its operator and destroyed a communications tower after being prevented from completing its mission. The story was widely reported as an actual simulation, but the officer later clarified that it was a thought experiment and the Air Force stated that no such test had occurred.<ref>{{Web citation|title=US air force denies running simulation in which AI drone 'killed' operator|date=2 June 2023|url=https://www.theguardian.com/us-news/2023/jun/01/us-military-drone-ai-killed-operator-simulated-test|retrieved=1 August 2026|website=The Guardian}}</ref> | | In 2023, a United States Air Force officer described a hypothetical scenario in which an AI-controlled drone attacked its operator and destroyed a communications tower after being prevented from completing its mission. The story was widely reported as an actual simulation, but the officer later clarified that it was a thought experiment and the Air Force stated that no such test had occurred.<ref>{{Web citation|title=US air force denies running simulation in which AI drone 'killed' operator|date=2 June 2023|url=https://www.theguardian.com/us-news/2023/jun/01/us-military-drone-ai-killed-operator-simulated-test|retrieved=1 August 2026|website=The Guardian}}</ref> |
|
| |
|
| The correction does not eliminate broader concerns surrounding autonomous weapons. Critics argue that delegating targeting or lethal decisions to automated systems can reduce accountability, accelerate warfare, reproduce faulty data, and make violence easier to authorize.
| | == Deployment by country == |
| | |
| A socialist approach opposes the use of AI for imperialist warfare, occupation, automated targeting, and political repression while distinguishing such applications from civilian, scientific, educational, or assistive uses.
| |
|
| |
|
| == Artificial intelligence in China == | | === China === |
| The [[People's Republic of China]] has incorporated artificial intelligence into national development planning. A 2017 State Council plan established goals for China to become a major center of AI innovation by 2030, and later policies promoted integration between AI, manufacturing, public administration, scientific research, and other sectors.<ref>{{Web citation|title=China accelerates AI development to build AI innovation center|date=6 April 2024|url=https://english.www.gov.cn/news/202404/06/content_WS6610834dc6d0868f4e8e5c57.html|retrieved=1 August 2026|website=The State Council of the People's Republic of China}}</ref> | | The [[People's Republic of China]] has incorporated artificial intelligence into national development planning. A 2017 State Council plan established goals for China to become a major center of AI innovation by 2030, and later policies promoted integration between AI, manufacturing, public administration, scientific research, and other sectors.<ref>{{Web citation|title=China accelerates AI development to build AI innovation center|date=6 April 2024|url=https://english.www.gov.cn/news/202404/06/content_WS6610834dc6d0868f4e8e5c57.html|retrieved=1 August 2026|website=The State Council of the People's Republic of China}}</ref> |
|
| |
|
| China combines state planning and regulation with substantial activity by state-owned, publicly influenced, and privately operated enterprises. Its AI sector therefore cannot be described as either wholly public or wholly private.
| | China's ''Interim Measures for the Management of Generative Artificial Intelligence Services'' took effect in August 2023. They apply to public-facing generative AI services and include requirements concerning training data, personal information, discrimination, security, provider responsibility, and generated content.<ref>{{Web citation|title=Interim Measures for the Management of Generative Artificial Intelligence Services|date=13 July 2023|url=https://www.chinalawtranslate.com/en/generative-ai-interim/|retrieved=1 August 2026|website=China Law Translate}}</ref> |
| | |
| === Regulation ===
| |
| China's Interim Measures for the Management of Generative Artificial Intelligence Services took effect in August 2023. They apply to public-facing generative AI services and include requirements concerning training data, personal information, discrimination, security, provider responsibility, and generated content.<ref>{{Web citation|title=Interim Measures for the Management of Generative Artificial Intelligence Services|date=13 July 2023|url=https://www.chinalawtranslate.com/en/generative-ai-interim/|retrieved=1 August 2026|website=China Law Translate}}</ref> | |
| | |
| The measures state that the government supports innovation while also requiring services to comply with Chinese law and uphold what the regulations call core socialist values. Supporters argue that this approach demonstrates greater state capacity to regulate corporations and direct technological development. Critics argue that the same framework enables state censorship and political control over information.
| |
| | |
| === Comparison with the United States ===
| |
| The United States AI industry is dominated by large private technology corporations, cloud providers, semiconductor firms, and investors. Federal policy has emphasized private-sector innovation, economic competition, military capacity, infrastructure expansion, and international technological leadership.
| |
|
| |
|
| | === United States === |
| In January 2025, the United States government issued an executive order directing agencies to remove barriers to American AI leadership. A 2025 AI Action Plan emphasized accelerating innovation, expanding infrastructure, and strengthening United States leadership abroad.<ref>{{Web citation|title=Removing Barriers to American Leadership in Artificial Intelligence|date=23 January 2025|url=https://www.whitehouse.gov/presidential-actions/2025/01/removing-barriers-to-american-leadership-in-artificial-intelligence/|retrieved=1 August 2026|website=The White House}}</ref><ref>{{Web citation|title=America's AI Action Plan|date=July 2025|url=https://www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf|retrieved=1 August 2026|website=The White House}}</ref> | | In January 2025, the United States government issued an executive order directing agencies to remove barriers to American AI leadership. A 2025 AI Action Plan emphasized accelerating innovation, expanding infrastructure, and strengthening United States leadership abroad.<ref>{{Web citation|title=Removing Barriers to American Leadership in Artificial Intelligence|date=23 January 2025|url=https://www.whitehouse.gov/presidential-actions/2025/01/removing-barriers-to-american-leadership-in-artificial-intelligence/|retrieved=1 August 2026|website=The White House}}</ref><ref>{{Web citation|title=America's AI Action Plan|date=July 2025|url=https://www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf|retrieved=1 August 2026|website=The White House}}</ref> |
|
| |
| The contrast between China and the United States is not simply regulation versus no regulation. Both states promote AI development and use AI for economic, administrative, and military purposes. China generally employs more direct state planning and content regulation, while the United States relies more heavily upon privately owned corporations and market competition.
| |
|
| |
| From a techno-socialist perspective, neither state policy alone guarantees democratic worker control. The relevant questions include ownership, public accountability, labor rights, surveillance, access, and whether productivity gains are socially distributed.
| |
|
| |
| == Artificial intelligence and the Global South ==
| |
| AI development is unevenly distributed internationally. Many states in the [[Global South]] depend upon foreign-owned cloud platforms, semiconductor supply chains, proprietary software, and externally controlled datasets.
| |
|
| |
| Supporters of technological sovereignty argue that developing countries require the capacity to build, regulate, adapt, and govern digital infrastructure according to their own social and economic priorities. Without such capacity, AI may reproduce dependency, extract local data, marginalize non-European languages, and establish new forms of [[digital colonialism]].
| |
|
| |
| AI applications may also support translation, education, agriculture, healthcare, public administration, scientific research, communication, and small-scale production when appropriate infrastructure and public institutions exist.
| |
|
| |
| Restrictions designed without public alternatives may preserve elite advantages when corporations and powerful state agencies retain access while ordinary users, small organizations, and poorer countries are excluded. Compliance costs, intellectual-property barriers, and restrictions upon computing resources may strengthen the largest firms.
| |
|
| |
| A socialist program for technological sovereignty would include public infrastructure, local language development, technical education, open standards, regional cooperation, democratic data governance, and resistance to foreign monopoly control.
| |
|
| |
| == Political communication ==
| |
|
| |
| === All Eyes on Rafah ===
| |
| In May 2024, an apparently AI-generated image bearing the slogan “All Eyes on Rafah” circulated widely on social media during the Israeli assault on [[Rafah]] in the [[Gaza Strip]]. The image showed rows of tents arranged around the slogan and was shared more than 46 million times through Instagram's “Add Yours” feature.<ref>{{Web citation|title=What is ‘All eyes on Rafah’? Decoding a viral social trend on Israel’s war|date=29 May 2024|url=https://www.aljazeera.com/news/2024/5/29/what-is-all-eyes-on-rafah-decoding-the-latest-viral-social-trend|retrieved=1 August 2026|website=Al Jazeera}}</ref>
| |
|
| |
| Supporters argued that the simplified image helped spread attention to the assault and allowed people to express solidarity with Palestinians. Critics argued that its sanitized appearance obscured the actual destruction and suffering in Rafah and displaced more direct forms of political action.
| |
|
| |
| The image demonstrated that AI-generated media could be used in pro-Palestinian political communication and complicated claims that support for generative AI is inherently incompatible with anti-imperialist politics.
| |
|
| |
| === Propaganda and misinformation ===
| |
| AI can generate false images, audio, video, and text at low cost. Such media may be used for propaganda, fraud, harassment, or the fabrication of evidence.
| |
|
| |
| The existence of synthetic media does not eliminate the older political problems of propaganda, censorship, concentrated media ownership, and state disinformation. It increases the speed and scale at which deceptive material can be produced and therefore strengthens the need for verification, media literacy, transparent labeling, and accountable institutions.
| |
|
| |
| == Political positions ==
| |
|
| |
| === Corporate technological development ===
| |
| Capitalist firms generally develop AI to obtain profit, market power, proprietary data, intellectual property, military contracts, and control over infrastructure. Corporate AI tends toward closed systems, subscription fees, surveillance, patents, dependence upon cloud monopolies, and decision-making by shareholders and executives.
| |
|
| |
| This does not mean every corporate deployment produces identical effects. Some systems provide socially useful services or expand access. The contradiction lies in productive technologies developed through collective labor and knowledge being controlled as private property.
| |
|
| |
| === Techno-socialism ===
| |
| [[Techno-socialism]] is a political tendency advocating the development and use of advanced technology under socialist ownership, public planning, democratic governance, and worker control. Techno-socialists reject both unconditional support for corporate AI and the complete rejection of artificial intelligence as a productive technology.
| |
|
| |
| According to this position, many harms associated with AI—including employment displacement, surveillance, environmental damage, uncompensated data extraction, and the concentration of wealth—result from capitalist ownership and deployment rather than from computation in itself.
| |
|
| |
| Techno-socialists argue that AI infrastructure should be treated as a socially important means of production. Proposals associated with this position include:
| |
|
| |
| * public or cooperative ownership of major computing infrastructure;
| |
| * democratic oversight of high-risk automated systems;
| |
| * union participation in workplace deployment;
| |
| * shorter working hours without reduced pay;
| |
| * universal access to assistive technology;
| |
| * public funding for open and transparent models;
| |
| * privacy protections and restrictions on data brokerage;
| |
| * prohibitions against predictive policing and abusive biometric surveillance;
| |
| * environmental regulation of data centers and semiconductor supply chains;
| |
| * compensation and bargaining rights for creative workers;
| |
| * publicly accountable education and research systems; and
| |
| * the distribution of productivity gains among workers and the wider population.
| |
|
| |
| ==== Public artificial intelligence ====
| |
| Public AI refers to models, computing infrastructure, datasets, and services owned or governed by public institutions rather than private shareholders. Such systems could support libraries, schools, universities, hospitals, public administration, scientific research, translation, and accessibility.
| |
|
| |
| Public ownership alone does not guarantee democratic control. A state-owned system may still be bureaucratic, repressive, or unaccountable. Techno-socialists therefore emphasize worker, community, and user participation alongside public ownership.
| |
|
| |
| ==== Democratic data governance ====
| |
| Data used by AI systems may include personal information, public records, creative works, workplace activity, and collectively produced knowledge. Democratic data governance would establish limits upon collection, require transparency, protect privacy, and give affected communities a role in deciding how data is used.
| |
|
| |
| ==== Automation and reduced working time ====
| |
| Techno-socialists argue that productivity gains should reduce the amount of necessary labor rather than produce unemployment and insecurity. Shorter working hours without reduced pay, strong public services, employment guarantees, and social ownership would allow automation to increase freedom instead of strengthening bosses.
| |
|
| |
| === Anti-AI movements ===
| |
| Opposition to artificial intelligence includes several distinct positions. Labor-oriented critics oppose job displacement, workplace surveillance, and uncompensated replicas. Environmental critics emphasize energy, water, mining, and electronic waste. Artists object to dataset collection, imitation, and corporate exploitation. Other opponents reject AI-generated media on cultural, moral, or philosophical grounds.
| |
|
| |
| These positions should not be treated as identical. Criticism of particular applications is compatible with support for other uses.
| |
|
| |
| ==== Criticism of technological rejection ====
| |
| Techno-socialists criticize blanket technological rejection for treating a means of production as though it possesses an independent class interest. Corporations, militaries, police agencies, and wealthy states are unlikely to abandon AI because left-wing movements refuse to understand or use it.
| |
|
| |
| Voluntary abandonment may instead concentrate knowledge and infrastructure among institutions already possessing capital. A technology cannot be democratized when ordinary people are excluded from understanding, governing, or accessing it.
| |
|
| |
| ==== Culture war and class politics ====
| |
| The division between “pro-AI” and “anti-AI” positions does not correspond directly to a division between social classes. Workers, corporations, governments, unions, artists, disabled people, educators, and political organizations contain differing views toward particular AI applications.
| |
|
| |
| Debates become cultural conflicts when they focus primarily on whether individual users are authentic, lazy, creative, moral, or socially acceptable. A class analysis instead examines ownership, employment, compensation, infrastructure, data control, working conditions, and the distribution of productivity gains.
| |
|
| |
| An individual worker using AI for translation, accessibility, study, art, or administrative labor does not necessarily share the class interest of a corporation using automation to eliminate employment or weaken a union. Similarly, opposition to AI is not automatically socialist, as it may also reflect professional gatekeeping, defense of corporate intellectual-property monopolies, brand protection, or cultural conservatism.
| |
|
| |
| === Consumer guilt ===
| |
| Individual consumption choices can have real consequences, but consumer shaming does not transfer ownership, strengthen unions, regulate surveillance, or transform energy infrastructure.
| |
|
| |
| From a Marxist perspective, political attention should focus on the corporations, states, investors, supply chains, and institutions that determine the scale and purpose of AI development. Individual use should be evaluated according to its actual effects rather than treated as the primary cause of structural problems.
| |
|
| |
| == Historical comparisons ==
| |
| Socialist and anti-colonial movements have historically distinguished between productive technologies and the ruling classes controlling them. Modern infrastructure developed under capitalist or colonial systems has frequently been appropriated, nationalized, or redirected by revolutionary movements.
| |
|
| |
| These comparisons do not imply that earlier revolutionaries held positions on contemporary AI. They illustrate a broader socialist distinction between machinery and the social relations governing it.
| |
|
| |
| === Marx on machinery ===
| |
| [[Karl Marx]] did not describe machinery as an independent enemy possessing its own class interests. He analyzed how capital uses machinery to control labor, increase relative surplus value, displace workers, and transform the labor process.
| |
|
| |
| Marx also recognized that developed productive forces could reduce necessary labor time and create material conditions for greater human freedom if social wealth were no longer governed by private accumulation.<ref>{{Web citation|author=Karl Marx|title=Capital, Volume I, Chapter Fifteen: Machinery and Modern Industry|url=https://www.marxists.org/archive/marx/works/1867-c1/ch15.htm|retrieved=1 August 2026|website=Marxists Internet Archive}}</ref>
| |
|
| |
| === Lenin and electrification ===
| |
| The early Soviet government treated electrification, technical education, industrial development, and modern infrastructure as essential to socialist construction. [[Vladimir Lenin]] did not regard machinery as permanently capitalist merely because it had developed within capitalist society.
| |
|
| |
| Techno-socialists compare contemporary computing infrastructure with electrification as a socially transformative technology whose political effects depend upon ownership, planning, and access. This comparison does not imply that Lenin held a position on modern artificial intelligence.
| |
|
| |
| === Revolutionary China and Cuba ===
| |
| The revolutionary governments associated with [[Mao Zedong]] in China and [[Fidel Castro]] in Cuba did not reject industrial machinery, transportation networks, hospitals, communications systems, or scientific institutions because they had previously operated under capitalist or colonial conditions. They sought to transform their ownership and social function.
| |
|
| |
| Supporters of techno-socialism use these histories to argue that socialists should seek control over new means of production rather than abandon them to corporations, militaries, and wealthy states.
| |
|
| |
| === Fanon and revolutionary radio ===
| |
| In ''[[A Dying Colonialism]]'', [[Frantz Fanon]] described the changing social role of radio during the [[Algerian Revolution]]. Radio had initially been associated with French colonial authority, European settler society, and colonial propaganda. During the liberation struggle, Algerians increasingly adopted radio to follow political developments and receive information connected with the [[National Liberation Front (Algeria)|National Liberation Front]].<ref>{{Web citation|author=Frantz Fanon|title=This is the Voice of Algeria|url=https://www.campusincamps.ps/wp-content/uploads/2015/10/fanon-this-is-the-voice-of-algeria.pdf|retrieved=1 August 2026|website=Campus in Camps}}</ref>
| |
|
| |
| Fanon's account demonstrated that a technology introduced under colonial conditions could acquire a different political meaning when appropriated by the colonized population. Techno-socialists draw a comparable distinction between technologies developed under capitalism and the social relations governing their later use.
| |
|
| |
| The phrase “by any means necessary,” associated with [[Malcolm X]], is also sometimes invoked by contemporary activists arguing that oppressed groups should not voluntarily surrender useful communications or productive technologies to their opponents. This represents a modern analogy rather than a documented position held by Malcolm X concerning artificial intelligence.
| |
|
| |
| == Corporate positions ==
| |
| Corporations may promote, restrict, or oppose particular AI applications according to profit, brand identity, copyright, workforce control, or market position. A corporation's opposition to one use of AI does not necessarily mean it opposes automation throughout its operations.
| |
|
| |
| === Nintendo ===
| |
| Nintendo has expressed caution regarding generative AI, particularly concerning intellectual-property issues and the preservation of the company's accumulated creative knowledge. Its position has not amounted to a rejection of every form of AI or machine learning.
| |
|
| |
| === Dove ===
| |
| Dove announced that it would not use AI-generated imagery to represent “real women” in its advertising. The policy concerns a specific form of brand representation rather than every use of AI by Dove or its parent company, Unilever.<ref>{{Web citation|title=Keep Beauty Real|url=https://www.dove.com/us/en/campaigns/purpose/keep-beauty-real.html|retrieved=1 August 2026|website=Dove}}</ref>
| |
|
| |
| === Selective corporate opposition ===
| |
| A company may reject generated advertising images while using machine learning for logistics, pricing, fraud detection, customer profiling, workplace monitoring, or labor reduction.
| |
|
| |
| Corporate anti-AI language should therefore be evaluated according to the material interests it protects. A firm may invoke ethics while defending proprietary software, copyright holdings, professional scarcity, or an established market position.
| |
|
| |
| == Regulation ==
| |
| AI regulation may address workplace displacement, discrimination, surveillance, privacy, automated public decisions, medical safety, military targeting, fraud, environmental effects, deceptive media, and the concentration of computing infrastructure.
| |
|
| |
| Labor-oriented proposals include:
| |
|
| |
| * requiring consultation with workers and unions before workplace deployment;
| |
| * prohibiting uncompensated digital replicas;
| |
| * protecting workers from algorithmic surveillance;
| |
| * requiring human review of consequential decisions;
| |
| * establishing legal liability for harmful systems;
| |
| * publicly auditing high-risk models;
| |
| * enforcing privacy and data-minimization rules;
| |
| * regulating data-center energy and water consumption;
| |
| * preventing monopoly control of essential computing infrastructure;
| |
| * supporting public and cooperative alternatives; and
| |
| * protecting the right of users to understand, repair, modify, or refuse automated systems.
| |
|
| |
| Not all AI applications present the same risks. Regulation may distinguish between assistive tools, consumer applications, workplace management, policing, military targeting, medical decisions, and critical infrastructure.
| |
|
| |
| A blanket ban is not equivalent to socialist control. Public ownership, democratic governance, worker power, universal access, and shared productivity gains constitute a different political program from either corporate deregulation or complete technological prohibition.
| |
|
| |
| == Criticism ==
| |
| Critics of techno-socialist approaches argue that describing AI as a neutral tool can understate how technologies are shaped during design. Training data, model architecture, funding, interfaces, energy requirements, and business models may embody the priorities of the institutions developing them.
| |
|
| |
| Others argue that public ownership does not automatically eliminate surveillance, censorship, bureaucratic control, or military use. State institutions can deploy AI repressively even without private shareholders.
| |
|
| |
| Techno-socialists respond that technology is neither socially neutral nor politically predetermined. AI systems are shaped by existing institutions, but those institutions can be contested and transformed. They argue that abandoning technological development would not remove it from society and would instead leave its direction to capital, militaries, and powerful states.
| |
|
| |
| The debate therefore concerns not only whether AI is beneficial or harmful, but which institutions possess the power to determine its development and whose interests it serves.
| |
|
| |
|
| == See also == | | == See also == |
|
| |
|
| * [[Automation]]
| |
| * [[Capitalism]]
| |
| * [[Communism]]
| |
| * [[Constant capital]]
| |
| * [[Cybernetics]] | | * [[Cybernetics]] |
| * [[Digital colonialism]]
| |
| * [[Imperialism]]
| |
| * [[Intellectual property]] | | * [[Intellectual property]] |
| * [[Karl Marx]] | | * [[Deepseek]] |
| * [[Means of production]]
| |
| * [[People's Republic of China]]
| |
| * [[Productive forces]]
| |
| * [[Scientific socialism]]
| |
| * [[Surveillance capitalism]]
| |
| * [[Techno-socialism]]
| |
| * [[Variable capital]]
| |
| * [[Essay:Intellectual property in the times of AI]] | | * [[Essay:Intellectual property in the times of AI]] |
|
| |
|
