The productivity paradox: why AI won’t speed up creative work

Light and Shade is a new series exploring the challenges at the heart of the AI-creative conversation. As AI becomes increasingly present across the creative industries, the series examines the opportunities and dilemmas our community grapples with. It is grounded in interviews with technologists, researchers, artists, designers, creative founders, writers, lecturers and environmental and computational experts, offering a fuller view of the many sides of the story of AI’s creative influence.

Throughout the generative AI boom, one particular promise has been repeated everywhere and endlessly, as if it were being whispered by the wind itself: It will save you time. AI tools will not replace your job, we’re told, but someone using them more than likely will. Why? Because they will have transformed into a super-fast, ultra-efficient, task-eating cyborg for whom no to-do list is insurmountable, all thanks to the help of their trusty machine-learning assistant. AI, said OpenAI CEO Sam Altman, “will act as a tool that enhances human productivity without compromising quality.”

For creative workers, the story is even rosier. Not only will AI save them time, but they will be able to redirect those rescued hours into the most stimulating aspects of their work. As a recent post on Adobe’s blog about agentic AI read: AI offers creative professionals “a pathway to growing their careers by freeing up time to do more of the things only they can do.” The tedious aspects of your work will be automated and, Google’s Sundar Pichai said in an interview with MIT, you will be liberated to “express yourself more creatively”.

It’s certainly a seductive vision. Liberated from routine drudgery, creatives can focus on the good stuff. And by achieving the same amount of work in fewer hours, they may even be able to work less, right?

There have been studies that suggest this AI-productivity hypothesis may check out. In 2024, researchers analysed the output of over 53,000 artists and 5,800 AI adopters on a major art-sharing platform and found that generative tools do, in fact, boost the number of works users post per month. Similar studies across other sectors have made comparable findings: everyone from customer support agents to programmers seem to complete tasks faster when using AI tools.

But is that how it’s actually playing out in real life? Are the creatives who have incorporated this technology into their working practices now luxuriating in an abundance of free hours that they can dedicate to meaningful tasks and general relaxation while their chosen AI does the hard yards? Are we in the midst of a renaissance? The anecdotal evidence, at least, suggests: No.

“Whoever thought AI would free creatives to be more creative wasn’t paying attention; it pushes us to produce more, not reflect more,” Sanchit Sawaria, a New York-based art director and design “generalist”, told us over Zoom. Sanchit has incorporated AI tools across his work, using them for experimentation and inspiration as well as more straightforward tasks like writing presentations. “I identify what I’m not good at and I use AI to do that for me,” he says.

Take Sorcerer for example, a digital musical instrument he created earlier this year. Despite having no background in coding, he used AI tools to create a program that produces sound by tracking hand movements through a camera. “I didn’t need to buy hardware, do deep coding, or pull open-source libraries for FM synthesis… I had a solid vision so I just described what I wanted, gave rough sketches and the AI tool helped turn it into code that I could then tweak,” he says.

But he’s yet to see any evidence that adopting AI tools will suddenly free up his time. “People now expect more and expect it faster. Because a video can be generated in 30 seconds, timelines for delivery have shrunk, but our mental capacity to sit with ideas, ponder and internalise has stayed the same.” 

This point of view – which troubles the official narrative being rolled out by AI companies – is not rare. In his newsletter “How to Survive the Internet” the author and journalist Jamie Bartlett paraphrased a recent conversation he’d had with someone he described as an “AI power user”: “‘So are you producing more stuff per day?’ I ask. ‘Absolutely. A lot more.’ ‘So do you have more free time?’ ‘Oh no – I’m busier than ever.’ ‘So are you earning lots more money then?’ ‘No – the same.’” Similarly, a recent Wall Street Journal headline read: “Your Prize for Saving Time at Work With AI: More Work”.

The artist, writer and technologist James Bridle – author of Ways of Being: Beyond Human Intelligence – told us that this paradoxical outcome should come as no surprise. “This is the history of labour for the last… however long, right?” they said. “The labour [argument] of 100-150 years ago was: ‘If we make work more efficient, we’ll all get to work less’. That argument was made in heavy industry, and it was made domestically too: if we have more appliances in the house, we’ll unlock more free time.” Of course, that never happens. “The market will always demand more, if there’s the capability of higher production,” says James.

Linda Dounia Rebeiz is an artist, designer and writer interested in the philosophical and environmental implications of technocapitalism. She uses AI tools throughout her practice: to research, code, write text, crunch data and generate images and videos. Some of her most recent projects, she says, would have been impossible without AI tools. In one, titled Once Upon a Garden, Linda spent more than three painstaking years using AI to carefully craft a speculative visual archive of extinct or critically endangered West African flora, for which there are little to no records. “I can do really crazy things now that I couldn’t do before,” she explains. “I can explore five different research questions at once, get synopses of each and use those to decide my next steps… But I’m not working less, I’m just working on more things at the same time.”

Has the use of AI tools allowed her to speed through the tedious parts of her work, freeing up time to focus on the more stimulating aspects? “Speed isn’t really the point,” she replies. “That discourse usually comes from people who don’t really use the tools in interesting or meaningful ways. There’s so much clickbait content, like: ‘How I set up an agent to handle my emails’. A lot of it doesn’t actually work well if you try it yourself. But it’s seductive content, shaped by social media and a very masculine productivity culture.”

If anything, Linda has noticed a counter-intuitive increase in drudgery as a result of using AI tools. “There’s setting up your tools, prompts, agents and so on. And then there’s so much maintenance – managing GPUs, dealing with overheating machines, clearing memory. You’re also generating a lot of junk content, so archiving becomes essential, and if you don’t name files properly or save progress consistently, then things get messy fast. That work is all tedious in its own way.”

Even if AI were the answer to eliminating tedium, Linda says you can’t just automate away anything mundane or repetitive in your creative workflow. “I actually think that’s bullshit. The grunt work is critical. Repetitive and tedious tasks put your brain into a different mode – one where deeper ideas can simmer. You can’t be ‘on’ all the time,” she says. “Creative breakthroughs take energy and intention, and they don’t come without the slow buildup. At least for me, I need the grunt work to get to the spark.”

A number of recent studies, while small in scale, have suggested that we grieve the loss of friction that comes from skipping over time-intensive or frustrating tasks, particularly in creative scenarios.

In 2024, researchers at Santa Clara University asked participants to generate ideas in response to creative prompts, allowing one set of subjects to use ChatGPT as an aid. The group that used AI generated more detailed ideas but felt less connected to them, and a common response from participants was that ChatGPT made the task fast and easy, yet mind-numbing. “ChatGPT allowed me to turn my brain off,” said one participant. Another commented that it “reduced the confidence I had to come up with creative things on my own.”

The problem is, while creative challenges might feel unending or laborious in the moment, they are, like most boring things, ultimately good for us. Judging which of your myriad of tasks are worth automating, and which are vital stops on the messy road to creative fulfilment, is going to become critical.

Another recent small study from MIT focused on the “cognitive cost” of using these tools by asking participants to write short essays, while wearing headsets that could monitor their brain activity. The participants that were allowed to use ChatGPT showed reduced brain activity, compared to those who only used a search engine or had no assistance at all. Over the course of the study, the ChatGPT users became increasingly reliant on the tool and, by the end, had resorted to copying-and-pasting its outputs into their essays, often struggling to remember anything they’d written. “The task was executed, and you could say that it was efficient and convenient,” Nataliya Kosmyna, the paper’s lead author, told TIME Magazine. “But as we show in the paper, you basically didn’t integrate any of it into your memory networks.”

The complicated and confusing reality behind AI’s supposed speediness is not only offset by increasing workloads, client demands and new maintenance tasks. There is a limitless feel to these AI tools, a sense that anyone can now make anything. Technical know-how is becoming vastly less important than simply having a good idea. Vibing a project – in which you simply speak the thing you want into existence by prompting – is now a popular approach. But there’s a feeling that this sudden sense of pure and unbridled potential can easily flip from a blessing to a curse. AI appears to offer limitless applications, yet as Orson Welles once warned: “The enemy of art is the absence of limitations.”

In a recent essay, the writer and designer Tina He wrote about how AI tools have led her into an endless cycle of escalating tasks and internal pressure, in which any time spent not working now feels like an unaffordable luxury. “As our tools amplify each hour’s potential yield, our internal expectations don’t just keep pace, they outrun our capabilities like shadows lengthening at sunset,” she wrote. She calls this “a psychological Jevons Paradox” – named after the 19th-century economist William Stanley Jevons, who observed that improved coal efficiency actually led to more coal consumption, not less. It’s a trap, she writes, “that threatens to consume our humanity in the pursuit of ever-greater output. We become victims of our own efficiency.”

Similarly, Sanchit Sawaria spoke about the slippery slope of not knowing when to stop when using AI tools. “It’s so easy to get the next iteration of whatever it is you’re trying to prompt. The ‘click a button, get a response’ paradigm encourages never stopping. But it doesn’t help you make decisions – it confuses you more,” he says. “I try to be mindful and intentional – if something isn’t working, I tweak the prompt specifically. But the UX design of AI tools encourages frictionless convenience. It tempts you to keep prompting without thinking deeply. That’s when you’re not directing well, you don’t really know what you want and it starts to feel like a slot machine.”

There will be ways for creatives to step around the potential traps of AI, but they will require a fundamental shift in how we think about work. It’s something of an open secret that the relationship between productivity and speed is a misnomer. More and faster usually equals worse. Generally, people accomplish the highest quality work when they slow down rather than speed up. Humans operate best at a humane pace. And while the mainstream discourse around AI continues to revolve around speed and productivity, it’s not surprising that the most interesting stuff seems to occur when creatives reject this framing entirely.

“Some of the things I do with AI feel like picking up a paintbrush for the first time,” says Linda. “It’s fun and often insightful.” She is now working with one other artist on a collaborative project called Pangool. Together, they are training a multimodal LLM (a large language model able to train on multiple types of data at once, such as text, audio, images, video and more) on the culture of the Serer people of West Africa, an ethnoreligious group primarily located in Senegal, The Gambia and Mauritania.

“Their entire culture is mediated through song, poetry and storytelling – it’s all oral,” she explains. “So you can’t train a standard LLM on that because there’s no written data, but we can with a multimodal LLM. That means it’s possible to represent and preserve an entire knowledge system. This multimodal model will become an explorable archive, and people around the world will be able to engage with Serer culture in a way that would never have been possible before.”

There is a version of the future in which human creativity and AI blossom, and the technology is incorporated in slow, spontaneous and thoughtful ways, as waves of new aesthetics, ideas and once-impossible projects are vibed into existence. And then there’s the other path, dominated by the time-obsessed cult of speed and delusions of efficiency: where workload expectations rocket, we offload more and more tasks to AI just to cope, output accelerates while skills atrophy, and we eventually reach a point at which we’re completely abstracted from the work that used to define us.

“There’s a big disconnect between what AI is marketed as doing and what it actually does well,” said Linda. “We risk repeating history. Like during the first industrial revolution, we thought machines would take over the drudgery and leave humans free for creativity. Instead, we just worked more. The incentives were misaligned; it became about shareholder profit, not better lives. AI isn’t just a production tool, it’s a thinking partner. But if we treat it like a mill, we’ll fall into the same trap.”