The broad implementation of artificial intelligence systems in the labor society will push the dynamic internal contradiction of capital to the extreme.
August 3, 2024
„The Internet? Is that thing still around?“
Homer Simpson, December 1999 [1]
The euphoria was followed by misery and disappointment. The introduction of the Internet at the turn of the millennium was accompanied by mass media hype and a speculative bonanza for high-tech companies (“Neuer Markt,” Nasdaq) on the stock markets, in which the internet industry was hailed as a new leading economic sector and millions of small investors started investing on the stock market (T-share, Infineon). For months on end, dubious IT companies were sometimes worth more than giant industrial groups such as Daimler. When the big high-tech bubble burst and a series of dubious IT start-ups went bust, a phase of disillusionment set in. The internet was ridiculed as a mere fad, merely a network for virtual sales outlets. And yet it cannot be denied that the internet industry fundamentally changed capitalism. Although no new leading economic sector in which the mass exploitation of wage labor would take place has been established, the former IT cliques that survived the massacre of 2000-2001 on the stock markets are now actually worth more than industrial corporations.
Link: https://exitinenglish.com/2024/08/03/ai-the-final-boost-to-automation/
Late capitalism is currently in a similar phase with regard to the economic potential of artificial intelligence. The great hype already seems to be dying down, as the first disappointing stock market results are being recorded – besides NVIDIA.[2] Furthermore, AI fatigue and disappointment are spreading among the public, as the grand visions of the AI gurus and transhumanists are still awaiting realization due to the clear shortcomings of the artificial learning systems used to date.[3] The internet boom at the turn of the millennium, similar to the waves of industrial rationalization in the last two decades of the 20th century, when industrial robots transformed Fordist assembly line production, also seems to confirm a central thesis of bourgeois economics: While new technologies may render masses of jobs obsolete, the same technological progress creates enough new occupational fields that – despite all the frictions –ensure the continued existence of the capitalist labor society.
MIT’s Technology Review, for example, recently propagated this thesis of the labor market’s ability to regenerate. Their article drew a wide arc over the crises and technological boosts that have occurred since the 1930s, when the question of whether “technological progress, through the increasing efficiency of our industrial process, is taking away jobs faster than it can create new ones” was also controversially discussed within the Roosevelt administration in the midst of the Great Depression.[4] In view of the development of the U.S. labor market in recent decades, where in 2018 around 60 percent of wage earners were employed in occupations that did not even exist before 1940, the Technology Review saw no signs of the adaptability of the labor market being outstripped by the rationalizing effects of automation. According to the Technoblatt, talk of the “end of work” is a “distraction” from the question of how artificial intelligence can be used to grow the economy and create new jobs. German trade union officials such as DGB head Yasmin Fahimi, who described a crisis in labor society triggered by increasing “digitalization” as “nonsense,” argue in a similar vein.[5]
A Look Under the Hood of The Valorization Machine
Labor is indeed the basis of capitalist society; according to Marx, it forms the substance of capital through its objectification in commodity-bodies during the production process. Labor creates commodity values. And it continues to be spent on a massive scale. The bare employment figures seem to support the arguments of the MIT magazine and the DGB, especially against the background of the current labor market situation in the U.S. and the FRG; after all, the officially calculated unemployment rate in the United States is particularly low.[6] In Germany, there is a shortage of skilled workers.[7]
However, this is, to a certain extent, a positivist view of things, which simply adds up the wage labor performed while failing to recognize the function of different types of labor, especially with regard to the valorization process of capital. The work offered for sale on the capitalist labor market must therefore be viewed in its overall social context in order to be able to make judgments about the stability of the labor society. Even if they appear profitable from a business perspective, not all forms of labor contribute to the valorization of capital in society as a whole. Capital is a totality that can only be understood in terms of its own social dynamics, which in its fetishistic irrationality is distinctly different from the narrow-minded, seemingly rational interest-based calculations of the market subjects (many on the left also have difficulties with this approach).
The Financial Times (FT) certainly knows how to differentiate when assessing the U.S. labor society.[8] In a negative summary of the neoliberal era written in 2023, the business journal criticized above all the formation of a service society, in which employment in the service sector rose from 45% in the 1970s to more than 60% in the second decade of the 21st century. At the same time, the proportion of workers in industry and the construction sector has fallen from 55% to less than 40%. According to the FT, the U.S. has been overtaken by China in terms of industrial production. Why is this a problem? Deindustrialization has been a key factor in the recent strategic economic policy paradigm shift in Washington, in which neoliberal free trade has been replaced by increasing protectionism.
From an economic perspective, all work is not created equal, as American technology magazines and German trade unionists imply in their milquetoast calculations. The commodity-producing industries form the “foundation,” so to speak, of the capitalist labor society. It is only on top of this that a service sector and a financial superstructure can be built – specifically in the form of wages and taxes. The welfare state, the education and care of future or former wage earners, and the maintenance of infrastructure must also be withdrawn from the capital valorization process as economic costs, even if individual companies (private kindergartens, universities, construction companies or retirement homes) profit from this on a business level. After all, not all wage earners can become hairdressers, financial managers, civil servants or waiters if there is no broad valorization of labor in industry.
A service society dominated by the financial sector, such as the deindustrialized, “rust belt” covered U.S. until the great real estate crash of 2008, can only reproduce itself by means of debt and bubbles until the inevitable crash. This is the lesson from the real estate crisis, as discussed by the FT, which led Washington to take a major protectionist turn. The role model is now Germany, which has been able to maintain its industrial base in the era of globalization through enormous export surpluses and a beggar-thy-neighbor policy (and this is precisely why Germany’s export industry is increasingly suffering from American protectionism).
Without a broad employment base in industrial production, there is no stable labor society – this is the conclusion from the era of neoliberal financialization and globalization, in which Marx’s concept of value, which distils the value of a commodity to the quanta of socially necessary labor time spent in its production, is also confirmed. Marx spoke of productive and unproductive labor with regard to the process of capital accumulation in society as a whole. Productive labor contributes directly to the valorization of capital within commodity production, while unproductive labor – as useful and socially necessary as it may be – does not do so directly. The crisis of labor society must therefore be seen as a crisis of productive, value-creating labor in industrial commodity production. The crisis of labor society is a crisis of productive labor, understood in the Marxian sense, meaning only the labor that contributes directly to the valorizing movement of capital.
This trend towards a shrinking industrial workforce, which has been lamented by the Financial Times based on what has happened in the U.S., can be empirically proven in almost all Western “industrialized countries.” Even in the export-oriented Federal Republic of Germany, which still has the strongest industrial sector in Europe, the proportion of people employed in manufacturing fell from just under 50% at the beginning of the 1970s to around 23% in 2023 as a result of automation in industrial production – while at the same time German industrial goods, such as machines and cars, flooded half the world.[9] The industrial foundation of capitalist labor societies is thus becoming increasingly fragile.
What’s more, with the onset of the third industrial revolution in the late 70s and 80s, which led to the major push toward automation in industrial production, total global debt rose faster than global economic output.[10] The late capitalist world system is thus increasingly running on credit; this debt creates demand for the sale of commodities, leading to a situation where many of the industrial jobs that still exist are simply dependent on demand generated by credit. The late capitalist world system is thus increasingly dependent on debt. However, this debt dynamic cannot be maintained for much longer in the face of the increasing distortions in the financial sphere and stubborn inflation. The illusion of an intact capitalist labor society, which German trade unionists and American technology magazines like to indulge in, can only be maintained by ignoring the conditions in the periphery of the world system – from whose collapsing regions and failed states economically superfluous wage earners are desperately trying to flee to the core.
A look under the hood of the capitalist valorization machine thus makes it clear that the optimism spread by American technology journals and German trade union officials on the eve of the great AI rationalization push is misplaced. Not only is the capitalist labor society gripped by a progressive erosion process in which its industrial base continues to melt away – the methods of delaying the crisis, in which this deficit-ridden zombie system produces ever greater mountains of debt, are also reaching their limits due to increasing instability in the financial sphere and stubborn inflation. The internal, moving contradiction of capital, which is getting rid of its own substance, wage labor, through rationalization, can therefore no longer be intercepted by these methods of credit-financed crisis delay during the next wave of automation.
In addition, a lack of investment in the welfare state, education and infrastructure increases the susceptibility of late capitalist societies to crises. This reflects the increasing imbalance between productive labor (the valorization of capital in the production of commodities) and unproductive labor (necessary expenditures on social infrastructure and the welfare state) in society as a whole. The particularities of the situation of Germany, which are often used to trivialize the crisis of the working society, do not change this. The universally lamented shortage of skilled workers and the ageing of society are precisely due to the fact that the dwindling share of wage labor performed in the production of commodities is offset by ever greater expenditure on the “dead costs” of social infrastructure (education, care, the welfare state, children seen as career killers and cost factors, etc.).
Office Workhorses Threatened with Extinction
In contrast to German trade union officials, who are probably plagued by a kind of job anxiety in this discussion, U.S. investment banks are certainly addressing the “disruptive” potential effects of the AI revolution on the global labor market. In a study published in mid-2023, Goldman Sachs estimated that “generative AI” (bots trained for specific work processes using mountains of data) will either downgrade or make obsolete around 300 million jobs worldwide. In a similar forecast published in early 2024, management consultants McKinsey concluded that in the United States alone, up to 30% of current working hours could become redundant by 2030, with low-paid, simple office work, customer service and sales being particularly at risk.[11]
Accountants and office workers in administration are particularly vulnerable. The first major wave of automation in the course of the third industrial revolution of microelectronics and the IT industry hit the workforce in the late 70s and 80s – now it’s white-collar workers’ turn to face the same fate. Office work could soon become obsolete on a massive scale. The more schematic the procedure, the less individual leeway there is in the work process, the easier it will be to automate it using AI systems, which can be “trained” for these work processes based on gigantic amounts of data that large corporations and offices have access to, following the example of the “large language models.” The office worker, the entire class of white-collar workers that first emerged en masse in the first half of the 20th century, is now threatened with extinction.[12] This class of petty-bourgeois wage earners, whose emergence dashed the old Marxist hopes of a revolutionary subject, is itself in the process of dissolution. The white-collar worker thus appears to have a relevant social span of existence of only about 100 years.
Humanity will soon have the ability to collect and organize information automatically, especially since administrative systems have already been almost completely digitized. The coming big AI wave will therefore build on the groundwork of the digitization that has been taking place in offices since the early 1980s. And it will be relatively easy to implement, as the investment costs are relatively low. The computers and administrative programs can continue to run; only the people operating them will disappear. The costs for office space and other “life support systems” will also be, for the most part, eliminated. On the other hand, companies must make substantial investments in data centers in which trained AI systems perform the former office tasks with minimal personnel costs, but these expenses are still low compared to cost of the industrial rationalization wave that began in the 1980s. Back then, the whole Taylor system had to be replaced and entire assembly lines equipped with robotic systems, each of which could cost millions.
Compared to these efforts to rationalize industry, which have cost billions of euros and caused the proportion of industrial workers to continue to decline, the investments being made in digital infrastructures, which will make employees obsolete, are insignificant. It is much cheaper to automate the white-collar worker now than it was for the industrial worker. In this respect, the late capitalist tendency to work from home, in the home office, which became established primarily in the wake of the pandemic, also points to the beginning of the capitalist labor society in the early modern era. As part of the so-called putting-out system, wage labor crept into the homes and huts of tenants, small farmers and craftsmen, who received materials and tools from the early capitalist “contractors” for the home production of commodities, which they then bought up and offered for sale on the market. Now capital is gradually releasing its wage-dependent employees into home-based work before the coming AI automation push makes them completely obsolete. In a classic dialectical negation of negation, the historical final stage of capital thus once again reflects moments of its history of ascent to a higher stage of development.
Digital Day Laborers: The Chatbot Takes Over the Call Center
Call center workers tend to be precariously employed and miserably paid. They often work at home. So at first glance, the automation of the call center industry could appear to be a progressive process, in view of these poor working conditions. Unfortunately, the people affected, who only have their labor to sell on the market, have their very existence threatened by this process. The obsolescence of call center employees is no longer a dream of the future, but a reality. The Swedish payment service provider Klarna was able to cut around 700 jobs after the company used an AI bot developed by OpenAI to handle service requests.[13] According to the company’s management, the AI system handled standard tasks such as cancellations or refunds just as successfully as its human competitors.
According to Klarna, customer satisfaction has remained as high as it was with human interlocutors. The decisive advantages of the system are obvious: the system speaks 35 languages, it is used in 23 countries, it has no limits on working hours, no wage demands or trade unions. According to initial internal forecasts, the service savings from using AI should translate into a profit of more than $40 million. The OpenAI system has already handled two thirds of all chat queries in customer service accurately, saving customers a lot of time. The company now sees itself as an “AI-supported global payment network,” according to the company’s enthusiastic conclusion in February 2024. At the same time, the share prices of call center operators such as Teleperformance and Concentrix have plummeted on the stock markets.[14]
At first glance, it would therefore appear that this wave of rationalization is primarily affecting simple jobs that require a low level of technical qualification. Cashiers and cab drivers, for example, are acutely at risk. At the moment, however, the AI-related losses in the service sector still seem to be cushioned by the AI industry’s need for unskilled workers who are used in the pattern recognition of artificial neural networks, known as the “learning phase” (see AI and The Culture Industry). Hundreds of thousands of precariously employed people, especially in the periphery of the capitalist world system, are busy coding the gigantic data sets of the AI systems with “labels” (similar to captchas when logging in) for miserable wages in order to enable these systems’ pattern recognition in the first place. The neural network only “learns” what a bicycle is when countless images of bicycles are given the label bicycle – the more, the better. The same applies to words, videos, music, etc. Of course, this does not mean that the AI understands what a bicycle is, as it is a purely external relationship that is established here.
What is actually taking place in this pattern recognition managed by digital day laborers is a process of internalizing all the digitizable images of external reality into the neural networks of AI systems. It is a gigantic process of scanning the mere surface of reality, without being able to take into account its dynamic character, its having become and its contradictions. The outer social shell, the false manifestation of late capitalism, becomes the inner essence of the AI systems, which will be incapable of critical reflection in principle, even in the event of rapid technological development – for example through quantum computers. The key point here is that the digitization of the surface of life, the universe, and everything else will be completed at some point, insofar as this is possible in algorithmic systems that know no causalities and only work with correlations. Consequently, the need for this mindless “learning work,” in which click-workers distribute labels for causalities and images of reality, will collapse, while AI will be able to handle many complex tasks. And this, precisely because it is fundamentally incapable of critical reflection in the emerging era of brutal crisis management (see “AI and Crisis Management”).
The Automation of The Middle Class
Mass media opinion-making is already being partially automated. The BILD newspaper, Germany’s most influential tabloid, wants to counter its internet-related loss of circulation and reach with a restructuring announced in mid-2023, in which the old business model will be brought up to date using AI.[15] A third of the tabloid’s 18 local editorial offices will be closed and a “three-digit number” of employees will be made redundant, while large language models will take over many everyday tasks. The Springer publishing house said that it was getting rid of “products, projects and processes that would never be economically successful again.” Generative AI should “contribute to supporting the entire journalistic process,” so that – literally – “journalism creation” becomes the core task area, while journalistic production becomes a by-product.
In the future, the large language models will be used for the tedious research that so often gets in the way of the seasoned BILD editor’s journalistic “creations.” Layout design, social media tasks and search engine optimization (SEO) will also be added. The fact that the AI models can already handle many “creative” everyday tasks in media operation was demonstrated by a recent scandal involving the renowned sports magazine Sports Illustrated. They secretly used texts generated by generative artificial intelligence, to which fictitious authors were also assigned – on top of that, the portraits of the fictional sports journalists were also generated by the AI.[16] The technology website CNET also secretly published AI-generated content.
In fact, the profession of journalist is one of a whole range of well-paid middle-class jobs that are threatened by partial automation and devaluation, according to a study by AI company OpenAI.[17] In addition to journalists, writers, mathematicians, interpreters and programmers are also at risk of becoming obsolete. A good proportion of middle-class jobs will therefore at least be devalued. The same applies to lawyers, graphic designers, financial advisors, analysts and stock market traders,[18] as well as the media industry from film to video games.[19] Wherever large amounts of data and information have to be processed in order to reach clear conclusions – for example in the legal system and legal advice – large language models are already ready for use. Memorization is becoming obsolete. Financial advisors and market analysts operate with probabilities resulting from the processing of empirical market data, which can now also be done efficiently by AI systems.
The second pillar of the automation of middle-class jobs is the modification of the data material that the large language models have scanned, such as the creation of new images, graphics, videos, texts, books, etc. Many tasks in the advertising industry are likely to be eliminated. Here, the simple, superficial modification of existing material by AI coincides with the ideological tendencies and the business model of the late capitalist culture industry, which thrives on the constant aesthetic repetition of the same old thing, which makes this type of automatic generation of “content” particularly seductive (see “AI and The Culture Industry”). The creation of films, entertainment books, and video games is predestined to be largely automated. A large proportion of jobs in the culture industry are under threat – precisely because it produces ideologically standardized content that only reflects the surface of social reality.[20]
In addition, well-paid jobs in advertising and sales that require direct work with customers also appear to be disappearing in the medium term (so it’s not just call centers that are affected). The insurance industry, for example, is spending billions on automating its administration and extensive legal departments and developing chatbots to streamline insurance sales and customer service.[21] In the meantime, photos of claims are even being analyzed by AI in a test operation. However, the chatbots, which are to be unleashed on customers as artificial insurance representatives, are still in the “test phase,” according to Spiegel-Online, as they first have to learn the jargon of the industry.
The marketers of the 21st century, the influencers running rampant on YouTube, Instagram, TikTok & Co., who try to sell their audience shit without labeling it as simple advertising, have already begun to be automated. Meanwhile, “hyper-realistic” (Ars Technica) virtual models are about to break into the approximately $21 billion market for “content” on social media.[22] In December 2023, Ars Technica reported on a successful AI model that was able to accumulate a following of 200,000 internet users in order to sell product placements for around 1,000 dollars per post. Such bots not only have the advantage of being completely controllable, which makes them more stable given the escapades of famous influencers.[23] What’s more, the mouths for hire that are rampant on social media have themselves contributed to their obsolescence by standardizing their appearance and look, which is imposed by the requirements of search engine optimization (SEO) and must be followed in order to get the highest possible number of hits. The influencer is already a sterile advertising product, largely shaped by algorithms, and they are now ripe for automation.
The reports about the possibilities of automation for the middle class are usually accompanied by reassurance pills: automation could never completely replace the professions concerned – lawyers, journalists, programmers, designers, creative professionals, etc. The professions concerned could concentrate more on creative activities and decision-making, while AI would deal with the daily grind, the schematic tasks. Of course, these objections should be taken seriously, and they are likely to correctly predict the near future, in which journalists, lawyers, book authors, etc. will continue to exist. However, the resulting higher productivity will lead to a displacement of workers from the professions concerned. Market-mediated capitalist production by isolated competitive subjects will lead to stronger predatory competition on the labor market, so that here too only a smaller number of labor providers with higher productivity will survive. Market competition will thus ensure that only the most productive wage earners, freelancers or self-employed workers with the most favorable price-performance ratio will survive.
The AI Ghosts They Summoned: The Slow Death of The Programmer
The AI revolution thus also leads to a devaluation of the skills of the commodity of labor, which can suddenly only be offered for sale on the labor market at a fraction of its former value – a constant tendency of capital as a moving contradiction, which led to the outbreak of the desperate Silesian weavers’ revolts as early as the 19th century. The U.S. magazine New Yorker recently published an interview with a programmer who described from his own experience how this technologically induced devaluation process is taking place in his industry.[24] At the beginning of the 21st century, when the internet experienced its big breakthrough, web designers could still earn good money by creating homepages – but these activities have now been largely devalued by software that almost anyone can use.
The situation is similar with the new AI programming bots that are now commonplace in the industry. A superficial, quickly acquired level of knowledge is now sufficient to solve complex problems quickly and efficiently. The subject of the interview, who became a programmer during the IT boom when he could set his salary more or less at will, described the successes of an acquaintance who used an AI bot for programming. The amateur with a cursory knowledge of programming languages was able to solve even complex problems in his hobby projects faster than the highly paid software developer. The AI tool GPT-4 is not only good at solving “tricky” small problems, it also has the “qualities of an experienced software developer,” as it can suggest good solutions and development paths for projects from a “large knowledge base.”
Until now, the motto in the industry has been that qualifications, that lifelong learning is the best protection against obsolescence, but now he would advise his children against wanting to become software developers. The infinitely complex art of programming machines in abstract programming languages is giving way to the technical dialog between user and AI programming tool that the vast majority of computer users can learn to use.
In fact, software development is an important focus within the automation efforts of the AI revolution, as the self-programming machine represents the Holy Grail of transhumanism, so to speak. This high-tech ideology hatched in Silicon Valley sees humans as a mere transitional phenomenon that are to be replaced by a permanently self-perfecting artificial intelligence – the so-called singularity.[25] This dystopia could only succeed if the process of programming AI bots can be accomplished autonomously, or in other words, if the AI can write its own code.
AI and The Outer Barrier of Capital
However, the high-tech Taliban and AI gurus who want to rake in billions in profits from human obsolescence face another external barrier: the finite resources of planet Earth, which is in the midst of a manifest climate crisis. The AI industry is already consuming huge amounts of energy and water.
According to studies from 2022, information and communication technology was responsible for 2.1 to 3.9% of global greenhouse gas emissions, which is roughly equivalent to air traffic emissions.[26] Added to this is the electricity demand of AI systems, which is set to explode to up to 134 terawatt hours by 2027 – roughly equivalent to the energy consumption of the Netherlands. At the beginning of 2024, the International Energy Agency (IEA) published its estimates regarding the energy consumption of the crypto and AI sector, which together were already responsible for around two percent of global energy consumption in 2022, with this share set to double by 2026.[27]
Added to this is the high water consumption of the hot-running data centers, which require water cooling systems. The annual water consumption of neural networks is expected to explode to 6.6 billion cubic meters by 2027, which would equal the water consumption of Denmark. During a “conversation” with GPT-3, in which 10 to 50 questions are answered, around half a liter of water is evaporated. As a reminder, two billion people around the world do not have regular access to clean drinking water, and 771 million people on earth cannot even reliably meet their basic needs.[28]
In order to train Microsoft’s GPT-3 with its 175 billion artificial neurons for a new task using gigantic amounts of data, an estimated 700,000 liters of water evaporate during the cooling process.[29] The electricity consumption for a single “training session” is equivalent to the annual consumption of 130 American households.[30] The learning phase of the large language models is considered to be particularly energy-intensive, but everyday operation, such as queries, is also characterized by high computing and energy consumption. A simple query answered by a large language model consumes around 30 times as much energy as the typical google search.
Just because it is sheer madness to waste gigantic amounts of energy on artificial neural networks in a manifest climate crisis does not mean that this project will somehow be stopped. For one thing, the fetishistic dynamics of capital are blind to the ecological and social consequences of their valorization compulsion. The world is merely a transitory stage for turning money into more money. Moreover, for transhumanism and similar ideologies, it is in fact a race between the ecological decay of the foundations of human life and the formation of the “singularity” inheriting humanity, which would no longer be dependent on such trifles as an intact environment. The hope is to reach the singularity before the social and ecological collapse. “Can what is playing you make it to level 2?”, as the accelerationist Nick Land put it.[31] Thus, capitalist rationality turns out to be a sinister idolatry, especially in the cult of AI, in which humans and nature are slaughtered on the altar of capital blindly moving as an automatic subject, which would come into its own in the singularity.
[1] https://getyarn.io/yarn-clip/813709cb-ba6e-435c-a171-c5450ce60533
[2] https://www.wallstreet-online.de/nachricht/17892567-konkurrenz-waechst-adobe-enttaeuscht-schwachem-ausblick-ki-gewinne
[3] https://www.konicz.info/2017/11/15/kuenstliche-intelligenz-und-kapital/
[4] https://www.technologyreview.com/2024/01/27/1087041/technological-unemployment-elon-musk-jobs-ai/
[5] https://www.spiegel.de/karriere/kuenstliche-intelligenz-auf-dem-arbeitsmarkt-beschaeftigte-fuerchten-jobverlust-durch-ki-a-452166c9-26c9-4805-a0f2-07e894292080
[6] https://www.bls.gov/news.release/pdf/empsit.pdf
[7] https://www.verdi.de/themen/arbeit/++co++74debf86-472f-11ee-894c-001a4a160129
[8] https://www.ft.com/content/77faa249-0f88-4700-95d2-ecd7e9e745f9
[9] https://de.statista.com/statistik/daten/studie/275637/umfrage/anteil-der-wirtschaftsbereiche-an-der-gesamtbeschaeftigung-in-deutschland/
[10] https://www.imf.org/en/Blogs/Articles/2023/09/13/global-debt-is-returning-to-its-rising-trend
[11] https://www.businessinsider.com/jobs-at-risk-from-ai-replace-change-chatgpt-automation-study-2023-7?IR=T
[12] https://de.wikipedia.org/wiki/Die_Angestellten
[13] https://www.newsweek.com/klarna-artificial-intelligence-tool-takes-700-jobs-1874002
[14] https://www.derstandard.de/story/3000000209642/bei-klarna-kann-ki-schon-hunderte-mitarbeiter-ersetzen
[15] https://www.forschung-und-wissen.de/nachrichten/technik/bild-zeitung-ersetzt-redakteure-durch-kuenstliche-intelligenz-13377679
[16] https://www.golem.de/news/kuenstliche-intelligenz-sports-illustrated-nutzte-heimlich-ki-texte-von-fake-autoren-2311-179818.html
[17] https://www.zdf.de/nachrichten/wirtschaft/kuenstliche-intelligenz-ki-arbeitsplaetze-chatgbt-100.html ; https://arxiv.org/pdf/2303.10130.pdf
[18] https://www.businessinsider.com/chatgpt-jobs-at-risk-replacement-artificial-intelligence-ai-labor-trends-2023-02?IR=T#legal-industry-jobs-paralegals-legal-assistants-3
[19] https://www.konicz.info/2024/03/05/ki-und-kulturindustrie/
[20] https://arstechnica.com/information-technology/2024/02/i-just-dont-see-how-we-survive-tyler-perry-issues-hollywood-warning-over-ai-video-tech/
[21] https://www.spiegel.de/wirtschaft/ki-experiment-der-versicherungen-wenn-herr-kaiser-ploetzlich-ein-chatbot-ist-a-b50e7bf7-fc7e-4e65-a136-c8c3ab65caa5
[22] https://arstechnica.com/ai/2023/12/ai-created-virtual-influencers-are-stealing-business-from-humans/
[23] https://www.youtube.com/watch?v=KuTsTjFZf5M
[24] https://www.newyorker.com/magazine/2023/11/20/a-coder-considers-the-waning-days-of-the-craft
[25] https://www.konicz.info/2017/11/15/kuenstliche-intelligenz-und-kapital/
[26] https://www.sciencedirect.com/science/article/pii/S2666389921001884 ; https://www.nzz.ch/technologie/chat-gpt-vs-googeln-der-massive-stromverbrauch-der-ki-ist-ein-problem-ld.1774379
[27] https://www.vox.com/climate/2024/3/28/24111721/ai-uses-a-lot-of-energy-experts-expect-it-to-double-in-just-a-few-years
[28] https://www.fr.de/wirtschaft/ki-studie-strom-verbrauch-umwelt-klimawandel-energie-zr-92745772.html
[29] https://mindsquare.de/karriere-news/chatgpt/
[30] https://www.theverge.com/24066646/ai-electricity-energy-watts-generative-consumption
[31] http://www.ccru.net/swarm1/1_melt.htm
Originally published on konicz.info on 04/19/2024