One by one, capitalism has co-opted every unique characteristic of the workers it resents the most: artists. Artists make unconventional work at unconventional hours, for example, so capitalism has branded this ‘workplace flexibility’ to undercut standard hours and basic entitlements. Artists make agile leaps between different ways of working, for example, so capitalism has branded this the ‘portfolio career’ driving a ‘gig economy’ that denies secure tenure. Artists unite to protect their interests via copyright laws, for example, so capitalism has rebranded IP theft as ‘training’ for generative AI programs, undermining artists’ and journalists’ rights. Capitalism has successfully assimilated innovation, resourcefulness and intellectual property into highly profitable uncertainty, precarity and content. And now, in the age of workslop, capitalism expects to integrate that special, most remarkable characteristic that it’s never been able to harness: creativity itself. 

But is that even possible? Early signs are not encouraging – and yet capitalism’s certainly not letting failure slow it down. Hallucinated references, pretend legal precedents, ghoulish images, fake news, slopaganda. The quality of what AI generates is nowhere near market-ready – but instead of retooling, employers are doubling down. Having made substantial investments in assets for which even their own developers have no long-term plan, more and more employers are demanding that staff increase AI use. This recasts employees into the passive roles of cogs in poor-quality machines, rather than autonomous professionals creating valuable work. The highly educated people previously hailed as powering the knowledge economy are being reduced to shovelling coal into the furnace: dirty input, dirty output. This work is unrecognised, deskilling and exploitative – it’s pretty darn hot at the coalface, especially when your own agency is at risk. With AI replacing journalists, entry-level jobs vanishing across multiple industries, and news publishers unashamedly calling themselves ‘AI input companies’, the nature and value of work is rapidly transforming. The question of the future of work is no longer What does work look like in the age of AI? but What does work look like in the age of AI slop? Because that future is already here – and it’s looking pretty sloppy indeed.

In the middle of a housing crisis, AI developers are displacing land for homes in the same way that property developers once displaced arable land. Like everything the tech giants are getting away with, the consequences are long-term and increasingly irreversible: we are losing the space needed to house people, the community amenity needed for thriving neighbourhoods, and the water needed to sustain life. Just as these multinational companies wheedle through local zoning regulations to build their thirsty hulking datacentres, AI developers wheedle through local copyright laws to feed large language models premised on copyright theft. Plenty is being written about the changing workplace’s anti-democratic turn, notably by Yanis Varoufakis in Technofeudalism (2023) plus a more recent warning: in his 21 April 2026 piece for The Point, Varoufakis identifies ‘techlordism’ as ‘the ideological cover for colonising everything—human endeavour, state institutions, and Wall Street itself’ as Big Tech moves to ‘legitimise replacing fallible, recalcitrant humans with cloud capital in every realm, from medicine to poetry translation to raising children.’ Software corporations, already bigger than governments, are now colonising our land, our neighbourhoods and our workplaces, having already colonised our attention spans, our social connections, our mental health, our news and our democratic engagement – or rather, disengagement.

Why would any workplace mandate increased use of AI under these circumstances? At a meeting of all staff on 8 April 2026, Nine Publishing’s managing director Tory Maguire did exactly this. As reported by Daanyal Saeed for Crikey, Maguire told the workforce – which spans the former Fairfax newspapers The Sydney Morning Herald, The Age and Australian Financial Review – that 

the company’s board expected everyone at Nine to be using AI, that Nine’s publishing division was underperforming in its utilisation of the technology, and that management would be monitoring staff usage of AI. 

Slides presented during the meeting directed staff to implement ‘Nine’s AI strategy “[to] encompass the full publishing ecosystem”, with newsroom operations to use it for “production, curation (and) planning” and journalists to use it for “archive research (and) data mining”’. Saeed described this directive as only just falling short of asking journalists ‘to use AI to write articles or generate story ideas’, while at the same time offering the Italian conservative newspaper Il Foglioas an exemplar – the same paper that had recently claimed to be ‘the first in the world to produce an edition entirely through AI.’ No confidence was given to staff on security of employment: when asked this question directly, Saeed reports ‘Maguire told staff that Nine could not guarantee what work would look like in three years, let alone six months.’ 

Using a new tool makes a lot of sense when the tool makes good work easier; using that same tool makes little sense at all when it makes your work worse, then enlists you to clean up its mess. Two recent American studies – one published in The Wall Street Journal on 21 January 2026, the other a partnership between Stanford Social Media Lab with AI-driven workplace coaching platform BetterUp, reported in The Guardian on 15 April 2026 – surveyed thousands of white-collar workers and found that sloppy AI outputs are costing even more time than having done the work themselves. On first impression, what’s been generated comes across just fine: the text or images seem polished and grammatically sound. Once read or studied closely, however, the output is nonsensical, illogical and riddled with errors. In the experience of one copywriting team: ‘initial drafts were a breeze to create’ until they ‘had to spend more time rewriting, correcting errors and resolving disagreements between each other’s chatbots than if they had never used AI at all.’

The Stanford team has named this phenomenon workslop: ‘AI-generated content that looks good, but lacks substance’. Workslop is what’s produced by the Content Mindset that I first identified last year in Griffith Review 88: Culture Vultures. Long pre-dating the advent of AI, the Content Mindset treats all creative and intellectual work as generic matter that fills the spaces between ads. This transactional approach devalues creative practice and journalism by treating art and news as interchangeable ‘content’. Extended into the workplace, the Content Mindset produces workslop when staff are expected to produce documents with only the semblance of meaning.

Why would any employer do this? It’s classic sunk cost reasoning: the expense has been made, the workforce must now adjust. ‘Companies have spent billions on enterprise investment in generative AI’, writes Ramin Skibba in that Guardian piece on the workslop studies. ‘Some of them, like Block, Amazon, Dow, UPS, Pinterest and Target, have laid off human workers at the same time, attributing the cuts to AI’s potential productivity’ – productivity boosts that have not been seen. The 2025 MIT report The GenAI Divide: State of AI in Business finds that 95% of companies who’ve made workplace AI investments have yet to see any financial return: they describe this as the ‘95% failure rate for enterprise AI solutions’. However, the view from the boardroom as top-level balance sheets are scrutinised doesn’t tend to cast an eye across the workplace to understand how that AI investment is being implemented. Instead, what tends to be imagined is a high-tech factory floor with shiny new tools whose investment must yield profit. And if they’re not, it’s the worker who must try harder.

One news publisher has gone even further. Taking workslop and the Content Mindset to a grim extreme, News Corp CEO Robert Thomson has described the Murdoch empire as ‘essentially an input company’ for AI development. In a 2 March 2026 presentation at the Morgan Stanley Technology, Media and Telecom conference, Thomson described his thousands-strong global news workforce as an AI input ‘in the way that semiconductors are an input, in the way that datacentres are an input, in the way that energy is an input’. He then announced an AI content licensing deal with Meta, premised on the news quality of The Australian, The Times of London and Dow Jones being ‘reliable’ and therefore ‘hard to beat’ as ‘an input’. Going back to that boardroom, we recognise the language of ‘reliable supply’ as more at home in primary industries contexts, describing raw materials moving efficiently through supply chains. An indictment and not a description of journalism, this is an extraordinary way to represent the highly-skilled professionals guiding how we understand the world and engage with democracy.

Not only are workers expected to clean up workslop’s mess, they’re also expected to feed its churn. The labour of re-doing workslop is not only time-consuming, it’s deskilling and it’s exploitative. ‘Quality decreased significantly, time to produce a piece of content increased significantly and, most importantly, morale decreased’, said one of the workers in the Stanford study. Another, a doctor, complained of ‘a lot of editing labour, frustration and concerns about data security and patients receiving AI-assisted emails with errors’. Professionals with significant education and experience are wasting time, costing money and losing confidence as they’re expected to do their jobs as well as retool the tools. Rather than exercise critical and creative skills, workers are becoming sausage factory fodder. There is no slop too sloppy not to be coopted back into capitalism’s endless churn.

Who benefits from the workslopped workplace? Big Tech. Software corporations have long ago infiltrated white-collar workplaces with oligopoly or monopoly dominance. How many times has an AI feature appeared without warning in your word processor, email or browser? And how many different apps can you name? Count the majors on one hand; it’s no longer possible to work, play or connect with one another without them – and Big Tech is getting bigger by the day. AI systems chip manufacturer NVIDIA, the world’s most valuable company ever, achieved a global all-time-record-breaking market capitalisation of US$5 trillion (AU$7 trillion) in mid-October 2025. That’s bigger than the GDPs of India, Japan and the United Kingdom, 12 times the size of the Australian sharemarket, or around the same size as Germany, the world’s third largest economy. AI systems developers Alphabet (Google, YouTube, Android, Pixel, Gemini AI and more), Meta (Facebook, Instagram, WhatsApp, Meta AI and more) and Microsoft (Microsoft 365, LinkedIn, Skype, Activision Blizzard, Copilot AI and more) have market capitalisations of US$1.7-4 (AU$2.3-5.6) trillion according to companiesmarketcap.com. To put that in perspective, Australia’s GDP of US$2.1 trillion (AU$3 trillion) ranks 12th in the world according to April 2026 International Monetary Fund data. These software corporations are absolutely everywhere in our workplaces – and they’re much bigger than governments. 

The consequences of workslop compound the insidious impacts of Big Tech on the changing ways we work – and there’s no shortage of impacts. The degradation of online platforms – what Cory Doctorow famously identified as enshittification in a 2022 blog post on the then mass exodus from Twitter and Facebook – is a strategy that’s intentional, profit-maximising, and by now, entirely normalised. We’ve long since come to expect to pay more for ever-declining quality, whether it’s trashy social media algorithms, poorly trained call centre staff in offshore sweatshops, or shrinkflation, that irritating practice of making packaging sizes smaller while maintaining the same price. In a 25 April 2024 piece for Medium titled ‘The specific process by which Google enshittified its search’, Doctorow outlines said process in detail, explaining Google’s strategy ‘to decrease the quality of search’ so as to force users ‘to spend more time on Google before they found what they were looking for.’ Doctorow identifies enshittification as a ‘macroeconomic’ and ‘microeconomic phenomenon’ that impacts both boardroom power dynamics as well as ‘the regulatory environment for competition, privacy, labor, consumer protection and IP.’ When market dominance leaves little to no realistic competition and nowhere for people to go, there’s barely any pretence of quality.

Within our workplaces, the consequences are more dire. When News Corp cautioned staff that AI use would be monitored, left unspoken was the workplace reality of continuous surveillance. Creative, news and knowledge workers aren’t just providing workslop ‘input’; those same platforms provide employers with ever-more-targeted modes of monitoring how people work. Surveillance offers far greater motivation for employers than any guarantee of AI quality. Not only can hours and locations of work be monitored, as well as emails sent and websites visited, but in the age of AI employers can monitor how an employee approaches a task and even their facial expression, with some employers mandating wearable monitoring devices. In their October 2025 article for the Journal of Business Research titled ‘AI and employee wellbeing in the workplace’, Alena Valtonen et al highlight the unprecedented intrusion of AI on the working body:

While AI adoption potentially improves organizational outcomes (Enholm et al., 2022), it also transforms how work is carried out and experienced by employees (Demerouti, 2022, Zirar et al., 2023), and how the roles of humans and AI are coordinated, aligned, and integrated in organizations (Anthony et al., 2023, Ramaul et al., 2025). These substantial changes raise concerns about the impact of AI on employee wellbeing (Nazareno and Schiff, 2021). For example, the development of technologies such as Walmart’s patented “performance metric” bracelet and Amazon’s Halo, which enable the monitoring of an employee’s productivity, voice tone, and emotions, has prompted concerns that AI managers will assess performance and influence career advancement (Mantello et al., 2023, Mantello and Ho, 2023). 

The authors undertake extensive literature review as well as empirical study based in Finland to examine the impacts of AI. Beyond wellbeing, their conclusion addresses the ways that ‘AI alters work characteristics—such as task execution, workflows, and safety—which in turn shape how employees experience their jobs. Thus, to maximize the wellbeing benefits of AI adoption, organizations must strategically implement AI to improve the aspects of work that are most important to employees.’ This, of course, is not the approach taken in most workplaces; the response to demoralising workslop been to compound the burden on workers.

For employers who aren’t in it for the surveillance, there’s absolutely no room for good-faith bosses to negotiate with Big Tech on quality to redress workslop. Like insurance, mortgages and transport tickets, software is a service-based product that bars any and all negotiation on terms. Contemporary workplaces are shaped by tools of trade that it’s no longer possible to download and own. Instead, subscription services license their use on unilateral terms that aren’t fixed but can and do change at Big Tech’s whim. Ubiquitous check-boxes demanding our consent on terms we never read have incrementally impacted more and more of our rights; use the AI features on conferencing software like Zoom or Teams and the content of your meeting will be datamined. It’s get-what-you’re-given: re-tooling is not an option. By forcing employees to keep at it, employers are co-opted into accelerating enshittification. Welcome to crapitalism.

‘Workslop is an unintended consequence of the AI boom’, writes Skibba, and the emphasis is mine – but is it really unintended? Whose interests are served? Hallucinated references, pretend legal precedents, ghoulish images, fake news, slopaganda: these phenomena, by now commonplace, dissolve confidence in research, journalism and the law, driving political disengagement and undermining democracy. Today it’s commonly understood that the algorithms powering social media have their own specific political biases as well as reinforcing discriminatory gender and cultural stereotypes. So too ‘AI-systems deliver biased results’ according to UNESCO’s 19 January 2026 resource on ‘Artificial Intelligence: examples of ethical dilemmas’. Building on its 2021 Recommendation on the Ethics of Artificial Intelligence, the UNESCO resource advises that ‘bias should be avoided… in the development of algorithms, in the large data sets used for their learning, and in AI use for decision-making.’ How likely are AI developers to heed this call? Back in 2023 an open letter hosted by the Future of Life Institute, referenced in my Griffith Review 88 piece, implored every AI lab in the world to ‘immediately pause for at least 6 months the training of AI systems more powerful than GPT-4’ to avoid ‘profound risks to society and humanity.’ Unsurprisingly the pause was not enacted, despite the predicted risk of ‘loss of control of our civilization’. For corporations bigger than governments, this perceived control is not a power they were ever likely to cede. 

As that power continues to grow, workslop’s pervasive busy-work makes for a useful distraction as our best and brightest minds are co-opted into the deskilling work of cleaning up its mess. In a 16 June 2024 article for the BBC titled ‘AI took their jobs. Now they get paid to make it sound human’, Thomas Germain observes that copywriters are ‘finding themselves being asked to team up with the same robots that are stealing their jobs to give the algorithms a bit of humanity – a hidden army making AI seem better than it really is.’ When workslop is normalised, it’s not just workplace characteristics like ‘task execution, workflows, and safety’ that are impacted; it’s how creative work is understood. Workers tasked with using AI to generate something start by entering prompts into a text field and then pressing a button, with editing the likely next task. Completely obscured are all the ‘inputs’ – ‘in the way that semiconductors are an input, in the way that datacentres are an input, in the way that energy is an input’, and in the way that stolen intellectual property and undervalued human labour are inputs. The generative in generative AI suggests that images and text can be produced instantaneously; prompts are understood not as sophisticated instances of critical thinking but as unrefined questions; editing as just churning slop into sense. Each of these key terms, previously at home in a creative environment, are co-opted and emptied of meaning. 

The more this language is normalised, the more creative work is devalued as something that can – and indeed, should – be generated instantly and effortlessly. The Content Mindset had already forced our first step along this downward path by normalising self-exploitation: when creative workers began calling their own work content and identifying as content creators, they began to validate the reduction of creative and intellectual work into mere matter that fills a given void. Content is transactional, interchangeable and disposable; creative work is the basis of culture, knowledge and art. It takes time to write a novel, paint a portrait or choreograph a performance, just as it also takes time to draft a difficult email, prepare a report or write a policy. Each one requires a sophistication of critical thinking, an understanding of context and a sensitivity to tone. When poor in quality, a great deal is jeopardised. 

The impact of poor-quality generative AI on the workplace is multifariously detrimental to how creativity is respected. Firstly, it takes work that’s long been understood as complex, intricate and labour-intensive, and simulates it in very little time, devaluing creative process. Secondly, it simulates that work poorly which then requires additional low-level labour, devaluing creative work. Once we recognise that not even the pervasive phenomenon of workslop has halted AI’s displacement of much-needed creative and intellectual jobs, we soon begin to understand workslop as a feature, not a bug, of capitalism. Workslop is an intended consequence of the AI boom: Big Tech profits as the quality of creative work declines. Thirdly, it demands that additional de-skilling labour of the very workers who were engaged to do the work in the first place, devaluing creative workers. It’s not only workplace characteristics that are impacted by workslop, it’s those elements that uniquely characterise the work of artists. What does work look like in the age of AI slop? By devaluing creative process, creative work and creative workers, workslop actively enables capitalism’s mission to co-opt every one of the artist’s coveted characteristics, including creativity. 

How best to resist that sloppy future? It’s in the grim distinction between generative and creative that we can make a positive start: let’s not see generative replace creative in the way that content is replacing work. Just as the Content Mindset entrenched the word content as a catch-all for intellectual property and creative work – displacing art, literatureand journalism – we now risk generative displacing creative, perilously undermining respect for creativity and creative professions. Of course, just as artwork isn’t content, generative AI is not creative. Creativity is a complex phenomenon of imagination and experimentation, conception and composition, iteration and reiteration. In the UNESCO 2005 Convention on the Protection and Promotion of the Diversity of Cultural Expressions, creativity is described as ‘a dynamic expression of identity and values, driving cultural diversity, social development and economic growth’. Creating those ‘cultural expressions’ requires ‘freedom of thought, expression and information, as well as diversity of the media’. It also requires workplaces that treat human beings as human beings.

To function, generative AI demands something deceptively basic of its ‘human input’: entering a prompt. So how do you ask a question that doesn’t yield a sloppy answer? The answer, ironically, is in precisely what workslop seeks to displace: critical thinking capacity. Thousands of years of philosophy across continents and cultures teaches us that the ability to ask an intelligent, meaningful question is one of the most sophisticated of all human accomplishments – and this, of course, is exactly what art does. As any artist knows, the prompt – the vision, the direction, the set of constrains – is more than half the work. Employers reducing the work of their staff to writing AI prompts lack the critical thinking to recognise that they’re draining the world of the critical thinking required to write that prompt in the first place – and indeed, to obviate its necessity. The more that workplaces stay stuck in that vicious cycle, the sloppier the workslop. 

Art asks questions – and demands that we do the same. Art shifts our perspective, making new thinking possible. Asking questions is expansive. It opens up our world. As an organisational leader I’ve long focused our workplace culture by structuring in time for shared critical thinking. Our weekly Staff Salons, most notably at Regional Arts Victoria, Melbourne Fringe and Craft Victoria, train one another in the art of asking questions. In doing so, they foster workplaces of dignity, as well as sparking unexpected collaborations. This particular workplace flexibility, however, is not open to everyone – which makes it a vital component of leadership for those who do have the power to lead with humanity. Because it’s not just those workers forced to wear monitoring devices whose behaviours are being constrained. It hardly seems coincidental that the consequences of workslop are preventing employees from engaging in the kinds of spontaneous conversations that foster critical thinking, creative innovation, and the kinds of democratic engagement that strengthen their own workplace rights.

A creativity-based approach to workplace rights is also a vital counter to workslop. Given most employees are required to assign intellectual property rights to their employer upon its creation in the workplace, let’s see a positive duty imposed on employers to keep that work meaningful. While the right to meaningful work is championed more commonly in disability advocacy to promote dignity and overcome segregation, we can expect to see it championed more and more as workslop drains workplaces of meaning. The Queensland Government, for example, identifies meaningful work in its 2025 State of the Sector report as a key element towards the ‘creation of an engaged workforce’ where ‘the employee perceives the importance of their work, understands the impact of work in the achievement of organisational objectives and, in the case of public servants, understands their contribution to the community’. The right not to be treated as a ‘human input’ is one we need to start seeing articulated and championed at the new frontiers of workplace organising. We need to make sure that workplaces remain environments of serendipitous encounter that creates insight, innovation and confidence, as well as resilience, good mental health and joy.

Of all the unique artist characteristics that capitalism has sought to co-opt, let’s not see meaningfulness itself colonised. Creative, not generative, is the human intelligence that contemporary workplaces need most. For the good of humanity, we need workplaces that can think for themselves again.