AI is getting better at writing. Humans must get better at editing Our investigation suggests how Jul 30th 2026 WRITING IS HARD. For Franz Kafka, putting words on the page was a process of “unending torments”. George Orwell said it felt like “a long bout of some painful illness”. Hunter S. Thompson, a journalist, saw it as the opposite of sex: “only good when it’s over”. Now AI has come to the rescue. Large language models (LLMs) eagerly offer reams of text without any pain or torment. Bots can spew out emails (excellently), longer prose (passably) and poetry (poorly). To them, writer’s block is a 500-sheet pack of paper: any user of an LLM can produce as many words in a day as Kafka did in his life.
For human writers this is a blessing and a curse. If you claim to be an author but use AI, you look like a fraud—bot-busting tools like Pangram are good at spotting when someone has had an AI helping hand. Take the hundreds of people who have used LLMs to write applications for The Economist’s internships, or the government ministers who have submitted AI-generated works to our By Invitation column (it’s tempting, but we won’t name them). Our investigation of the stylistic quirks of the big bots reveals how, with every update, AI writing is getting closer to human writing. We compared the outputs of LLMs to prose we know is produced by humans: our own. We asked 14 variants of ChatGPT, Claude, Gemini and Grok released since 2024 to write versions of our articles without consulting the web. We compared our corpus across 55,940 sentences and 1.2m words, and checked our results against other newspapers and hit fiction books. We found that some of AI’s former tricks—using lots of em-dashes, for instance, or words like “leveraging”—are no longer leveraged much in its writing. It’s a good thing that AI writing is getting better. Sifting through today’s slop is a drain on your brain and your time. LLMs have tripled the number of new e-books published on Amazon every month, to 300,000, a recent study found, and draft more than a third of new websites by one count. If LLMs keep improving, in a few years they could be as good as professional wordsmiths. In the meantime, though, their baggy prose puts a premium on editing skills. After all, even the best writers need a good editor to keep them sharp. Editing is something The Economist deals with day in, day out, so here are our tips for re-writing AI prose. First, commission well. George R.R. Martin once said the job of editors is to “clip the wings” of writers, and “tell them in which direction to fly”. When it comes to LLMs, that means giving clear instructions about exactly what you want, and asking them to refine their output as many times as necessary. You’ll get less pushback from machines than men. Mr Martin said editors were “the writer’s natural enemy”; AI would probably say “great observation!” and “love the enemy analogy!” Second, pay attention to detail. AI writers will make mistakes and offer phrases that sound good but mean nothing. They like to lavish their prose
with Latinate words, talking about the “bi-directional flow” and how to “expedite grid enhancements”; remind them that Anglo-Saxon words are often punchier. They like clichéd metaphors: there are lots of “Achilles’ heels”. And AI tends to use the “rule of three” far more than humans. The Economist also likes this rhetorical trick. But whereas we try to group three sensible bits of advice, AI offers such redundant sentences as: “The mystery has been solved—not partially, not ambiguously, but definitively.” Such lines should be cut—definitively. Finally, know your audience. The editor is an intermediary between writer and reader. LLMs are sycophants and their smarmy tone is often off-key. If you are writing an email to break some bad news to your boss, tell your bot to skip the exclamation marks. AI can be pretentious or pontificating (it was trained on human writing, after all). Most people are short of time, but a verbose AI pins you to the wall like a bore at a party. “Write drunk, edit sober,” goes the old writers’ maxim. If that’s right, when the AI-human team gets down to work, the AIs will have all the fun. ■ This article was downloaded by zlibrary from https://www.economist.com/leaders/2026/07/30/ai-is-getting-better-at-writing- humans-must-get-better-at-editing
America’s startup boom offers hope for the future of work AI could make more people their own boss Jul 30th 2026 MANY AMERICANS long to be their own boss. Their yearning is as old as the Republic itself: Thomas Jefferson thought the ideal citizen was a yeoman farmer, whose economic independence allowed him to participate in civic life. For decades, 60% of Americans have said they want to be self- employed. Yet in recent years fewer than 10% have been. That is now changing. As we report, America is in the midst of a small- business boom. It began during the covid-19 pandemic and has since picked up steam, fuelled by artificial intelligence, low hiring and the growing ease of starting small online shops on platforms such as Etsy, Shopify and TikTok Shop. In June the Census Bureau recorded more than 531,000 new business
applications, a rise of 81% compared with the average month in 2019. Such factors have buoyed self-employment in Britain and France, according to Stripe, a payments firm. But it is America where the trend is most pronounced. It is a welcome one. Small businesses are popular with politicians, but economists can turn up their noses. Many fail. Those that survive often suffer from low productivity: small American firms are only 70% as productive as big ones. In recent years much innovation has come from large “superstar” firms. Small companies often lack the digital tools and administrative efficiency that come with scale. Yet there are signs that AI can help small businesses become more sophisticated, even as it automates away some jobs with big employers. Basic chatbots can help firms register with local authorities, draft business plans and market themselves to clients. AI agents can carry out repetitive administrative tasks and make sense of sales trends and other data. Many of America’s new businesses are one-man bands, categorised by the government as unlikely to hire within their first few years of operation. They may be able to scale up anyway, using AI agents instead of employees.
All this raises the tantalising prospect that self-employment could provide meaningful alternative work for some who lose their jobs to ai. Gloomy futurists in Silicon Valley warn that AI will create a “permanent underclass” of people who own little capital and, with the professional classes displaced, perform only menial jobs. A more hopeful possibility is that the technology democratises the ability to put ideas into practice and become a successful business owner. There are some signs of such an effect: a rising share of new entrepreneurs are low-income, according to Bank of America. True, becoming a “solopreneur” will provide neither the manufacturing jobs craved by populists nor a predictable 9-5 office life and a corporate ladder. The new startups are mainly services firms, many in tech, retail or health care. But, when asked, their founders say they are happy with service work. It is an encouraging precedent that many of these new businesses have sprung up in places formerly hollowed out by the loss of manufacturing. Shopify reckons that the share of new businesses from rural areas has been rising. The government could make life easier still for entrepreneurs. Although Obamacare, a health-care reform in the 2010s, helped, it is often still too hard for self-employed Americans to access health insurance. Employees should not have to cling to their jobs to stay covered. And swathes of the economy suffer from burdensome occupational-licensing requirements, which act as a barrier to entry for anyone looking to set up shop. They should be pared back. Nobody knows how AI will reshape work, but across-the-board pessimism is unwarranted. An economy powered by individual entrepreneurship, with AI handling tedious back-end tasks, could yet prove fulfilling and lucrative for many people. It might even help more Americans realise Jefferson’s vision of self-reliance as a path to liberty. ■ This article was downloaded by zlibrary from https://www.economist.com/leaders/2026/07/30/americas-startup-boom-offers-hope-for- the-future-of-work
Does Britain’s social housing subsidise the wrong people? Also this week, weather forecasts, using AI in law, falling cows, drones and zebras, questions Jul 30th 2026 Letters are welcome via email to letters@economist.comFind out more about how we process your letter As representatives of both social-housing residents and providers we were struck by the way your recent article described people who live in social housing (“The rent is too damn low”, July 11th). Phrases such as “subsidises the wrong people”, references to residents who “cling” to their homes and examples chosen to illustrate perceived unfairness all contribute to a familiar narrative: that social-housing residents on good incomes are recipients of undeserved subsidies.