What if the AI is the paperclip in the story, and we’re the ones obsessed with making them? CC-licensed photo by Rob Faulkner on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

---

A selection of 9 links for you. Holding together. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

---

Can AI chatbots write emotionally rich romance books? • The New York Times

Alexandra Alter:

»

Whenever the publishing industry is rocked by a technological shift, it usually hits romance first. Romance writers are prolific and their readers are voracious, so they’ve been early adopters of e-book subscription services, self-publishing, social media networking and online serial releases.

Romance is also the publishing industry’s best-selling genre. It accounts for more than 20% of all adult fiction print sales, according to Circana BookScan, and has continued to grow in recent years even as overall adult fiction sales have stagnated.

The genre may be especially vulnerable to disruption by A.I., for all the reasons that readers love it. Romance relies on familiar narrative formulas, like the guarantee of an “H.E.A.” or “happily ever after.” And romance novels are often built around popular plot tropes — like enemies-to-lovers or forced proximity — that can be fed into a chatbot.

AI remains contentious in the romance community. A vocal contingent of readers oppose its use and are quick to call out suspected transgressions. Furore erupted on social media last year when two romance authors published works with AI prompts accidentally left in. “You’re an opportunist hack using a theft machine,” the fantasy writer Rebecca Crunden wrote in an expletive-laced message on Bluesky.

Many readers seem to share her distaste, she said in an interview: “The comment I keep seeing is, ‘Why should we pay for something that you couldn’t be bothered to make?’”

But without such obvious slip-ups, it can be hard to spot AI-generated romance. Amazon asks authors who use its Kindle self-publishing platform to disclose if they relied on AI, but does not require writers to include any public disclaimers on their books.

“The AI detector can be gotten around,” said Christa Désir, vice president and editorial director of Bloom Books, a romance imprint that often signs successful self-published authors. “It will become undetectable at a certain point.”

«

Correction: it will become undetectable for bad or mediocre writing. That will mean that the really good writers will stand out because they’ll construct characters and plots that truly resonate. Though of course, a lot of people cannot tell the difference; sickly sweet suits them just fine.
unique link to this extract

---

Is AI the paperclip? • New Cartographies

Nicholas Carr:

»

In a paper published in 2003, the philosopher Nick Bostrom sketched out a thought experiment aimed at illustrating an existential risk that artificial intelligence might eventually pose to humanity. An advanced AI is given, by its human programmers, the objective of optimizing the production of paperclips.

The machine sets off in monomaniacal pursuit of the objective, its actions untempered by common sense or ethical considerations. The result, Bostrom wrote, is “a superintelligence whose top goal is the manufacturing of paperclips, with the consequence that it starts transforming first all of earth and then increasing portions of space into paperclip manufacturing facilities.” It destroys everything, including its programmers, in a mad rush to gather resources for paperclip production.

Bostrom went on to refine his “paperclip maximizer” thought experiment in subsequent writings and interviews, and it soon became a touchstone in debates about AI. Eminences as diverse as Stephen Hawking and Elon Musk would routinely bring it up in discussing the dangers of artificial intelligence. Others were skeptical. They found the story far-fetched, even by thought-experiment standards. It seemed, as The Economist wrote, a little too “silly” to be taken seriously.

I was long in the skeptic camp, but recently I’ve had a change of heart. Bostrom’s story, I would argue, becomes compelling when viewed not as a thought experiment but as a fable. It’s not really about AIs making paperclips. It’s about people making AIs. Look around. Are we not madly harvesting the world’s resources in a monomaniacal attempt to optimize artificial intelligence? Are we not trapped in an “AI maximizer” scenario?

«

Carr always brings a new perspective to an idea. And as he points out, Elon Musk isn’t satisfied with consuming all the space on the planet. He wants to fill the space above us with AI too.
unique link to this extract

---

AI took over the Super Bowl, accounting for 23% of ads • AdWeek

Trisha Ostwal:

»

The Super Bowl has long been a proving ground for new technologies seeking mass legitimacy. This year, generative AI took the field in force. But despite years of hype, ballooning valuations and rapidly growing usage, many of the ads leaned on familiar promises and vague positioning. Four years into the AI hypecycle and it’s never been more clear that AI is in a messaging crisis.

According to iSpot, 23% of Super Bowl commercials—15 out of the 66 ads—featured AI. This includes giants like OpenAI and Anthropic—selling AI directly to consumers—while consumer brands, including vodka maker Svedka, leaned on the technology to make its ad. Taken together, the ads reflected a broader AI arms race this football season, with brands across product categories framing the technology as inevitable, while promising smarter tools, more human interactions, and the seamless integration of AI into everyday life.

Yet much of these advertisements struggled to clearly articulate what sets one offering apart from another, even as AI becomes mainstream.

[But what about Claude’s “hit” adverts on ChatGPT getting adverts?] Early audience response suggested the message struggled to land. According to an iSpot survey of 500 viewers, the ad’s likeability score placed it in the bottom 3% compared with Super Bowl ads over the past five years. Its top-two-box purchase intent scored 24% below Super Bowl norms and 19% below ads in its category that aired over the last 90 days. Viewers most commonly described their reaction as “WTF,” signaling confusion around both the message and the execution.

“Claude’s ad has done a good job of stoking conversations and controversy online, but the ad does not test well among general population audiences with confusion around the message and execution,” said Sammi Scharninghausen, brand analyst at iSpot.

«

But everyone in Silicon Valley loved that it was making fun of ChatGPT! Turns out that’s not how the rest of the population – which tuned in for some songs, adverts and occasional bursts of gridiron football – saw it.

At least everyone will have forgotten in a year or so when Claude introduces ads.
unique link to this extract

---

Australia’s grid now relies on renewable energy as much as coal. Those who doubted it look foolish • The Guardian

Adam Morton, writing at the end of January (the height of Australia’s summer):

»

Australia’s power grid is changing rapidly – so rapidly that it can feel difficult to keep up.

This week [ending 30 January], as an oppressive heatwave in the country’s south-east rewrote temperature records, there was also plenty of evidence demonstrating just how fast long-held assumptions about the electricity system are being overturned.

A significant part of the change is due to the astonishing rise of solar power, and the extent to which it is squashing coal generation. The grid is now operating in a way that many people considered unimaginable, and maybe impossible, not that long ago.

Back then, some commentators claimed the grid would not be able to function with more than 10% – and definitely not more than 20% – electricity coming from solar and wind. Those predictions look foolish now.

Over the past seven days [to 30 January], solar provided 30% of all electricity in the country’s main grid, which supplies the five eastern states and the ACT. That’s across day and night.

If you narrow the calculation to consider just when the sun is out, the numbers are even more striking. Solar met 59% of electricity demand between 9am and 6pm. More than half of this – 37.6% of the total – was from small-scale systems spread across about 4m roofs. The rest was from large-scale solar farms.

Dylan McConnell, a senior research associate at the University of New South Wales, says between 12pm and 1pm solar output peaked at 67% of consumption. It was more than 70% in New South Wales and South Australia.

Coal-fired power, the historic backbone of the grid that once supplied nearly 90% of power, could not compete. Solar energy is incredibly cheap. It costs much more to burn coal. It meant the country’s ageing coal fleet was reduced to filling in gaps, kicking in barely a quarter of the electricity used over lunchtime.

«

Coal was Australia’s power source for electricity for decades. There’s a debate going on there about nuclear, but it looks as though solar plus batteries will make that argument pointless.
unique link to this extract

---

Eileen Gu, Olympian by day, millionaire social media star by night • Yahoo Sports

Dan Wolken:

»

On January 9, freestyle skier Eileen Gu posted on Instagram a “Day in my Life” video commemorating the one-month countdown to the Milan Cortina Olympics.

Glimpses of Gu brushing her teeth, eating breakfast, riding a ski lift, practicing tricks on a giant airbag, cooling down with a five kilometre run, talking to the media, completing a doping test, and reading a book while she’s in a hyperbaric chamber were all stuffed into a 29-second Instagram reel.

Within 10 days, it generated more than a million views.

So it’s probably no coincidence that Gu, the American-born dual gold medal winner who chose to compete for China in 2019, was the fourth highest-paid female athlete of 2025. According to Sportico, all but $20,000 of her $23m came from endorsements.

While sponsorships have always been crucial to the earning potential of Olympic athletes, who generally aren’t raking in huge sums of prize money, financial success no longer hinges on whose image lands on the Wheaties box.

Now most of it happens on social media, where the line between Olympic athletes and influencers has been blurred — usually to the benefit of their pocketbooks as this Winter Games draws close.

“The number of Olympians who have become more popular and made money in the Olympics has grown exponentially,” said Doug Shabelman, the CEO of Chicago-based Burns Entertainment, a firm that matches celebrities and athletes with marketing opportunities. “Now everybody, whether you win or lose, can be an influencer. Before you had to win gold medals, you had to do something special. Now social media has leveled the playing field, and the marketability of these athletes is 365 days a year.”

«

Also up there: Simone Biles (gymnastics) and Lindsey Vonn (skiing, now out). Definitely a better way of making money from sports that otherwise get little TV attention.
unique link to this extract

---

Why Apple’s iOS 26.4 Siri upgrade will be bigger than originally promised • MacRumors

Juli Clover:

»

The iOS 26.4 version of Siri won’t work like ChatGPT or Claude, but it will rely on large language models (LLMs) and has been updated from the ground up.

The next-generation version of Siri will use advanced large language models, similar to those used by ChatGPT, Claude, and Gemini. Apple isn’t implementing full chatbot interactions, but any upgrade is both better than what’s available now and long overdue.

Right now, Siri uses machine learning, but it doesn’t have the reasoning capabilities that LLM models impart. Siri relies on multiple task-specific models to complete a request, going from one step to another. Siri has to determine the intent of a request, pull out relevant information (a time, an event, a name, etc), and then use APIs or apps to complete the request. It’s not an all-in-one system.

In iOS 26.4, Siri will have an LLM core that everything else is built around. Instead of just translating voice to text and looking for keywords to execute on, Siri will actually understand the specifics of what a user is asking, and use reasoning to get it done.

Siri today is usually fine for simple tasks like setting a timer or alarm, sending a text message, toggling a smart home device on or off, answering a simple question, or controlling a device function, but it doesn’t understand anything more complicated, it can’t complete multi-step tasks, it can’t interpret wording that’s not in the structure it wants, it has no personal context, and it doesn’t support follow-up questions.

An LLM should solve most of those problems because Siri will have something akin to a brain.

«

Narrator’s voice: it isn’t “akin to a brain”. But being able to hold the answers to a list of more than two questions would be useful.
unique link to this extract

---

Why Efficiency (E) cores make Apple silicon fast • The Eclectic Light Company

Howard Oakley:

»

What makes the difference in Apple silicon Macs is how threads are allocated to the two different CPU core types on the basis of a metric known as Quality of Service, or QoS.

As with so much in today’s Macs, QoS has been around since OS X 10.10 Yosemite, six years before it became so central in performance. When all CPU cores are the same, it has limited usefulness over more traditional controls like Posix’s nice scheduling priority. All those background tasks still have to be completed, and giving them a lower priority only prolongs the time they take on the CPU cores, and the period in which the user’s apps are competing with them for CPU cycles.

With the experience gained from its iPhones and other devices, Apple’s engineers had a better solution for future Macs. In addition to providing priority-based queues, QoS makes a fundamental distinction between those threads run in the foreground, and those of the background. While foreground threads will be run on P [power] cores when they’re available, they can also be scheduled on E [efficiency] cores when necessary. But background threads aren’t normally allowed to run on P cores, even if they’re delayed by the load on the E cores they’re restricted to. We know this from our inability to promote existing background threads to run on P cores using St. Clair Software’s App Tamer and the command tool taskpolicy.

This is why, even if you sit and watch all those background processes loading the E cores immediately after starting up, leaving the P cores mostly idle, macOS won’t try running them on its P cores. If it did, even if you wanted it to, the distinction between foreground and background, P and E cores would start to fall apart, our apps would suffer as a consequence, and battery endurance would decline… Efficiency cores get the background threads off the cores we need for performance.

…For an M4 Pro, for example, high QoS threads running on the P cores benefit from frequencies close to the P core max of 4,512 MHz. Low QoS threads running on the E cores are run at frequencies close to idle, typically around 1,050 MHz. However, when the E cores run high QoS threads that have overflowed from the P cores, the E cores are normally run at around their maximum of 2,592 MHz. By my arithmetic, 1,050 divided by 4,512 is 0.233, which is slightly less than a quarter. Other M-series chips are similar.

«

unique link to this extract

---

UK electric vehicle charging firms “seeking buyers amid rising costs and tough competition” • The Guardian

Jasper Jolly:

»

British electric charger companies are asking rivals to buy them as they run out of cash amid rising costs and intense competition, according to industry bosses.

A wave of mergers and acquisitions is likely to shrink the number of charge point operators from as many as 150 to a market dominated by five or six players, said Asif Ghafoor, a co-founder of Be.EV, a charging company backed by an investment fund owned by Octopus Energy.

Investors rushed to pour money into green technologies and the electric car industry during the pandemic, fuelled by cheap borrowing. Yet now with intense competition, rising costs, and delays to government funding, some charger companies are running short of cash and investors are looking for a return on their investments, according to several people in the industry.

Simon Smith, the chief executive of Voltempo, which focuses on charge points for lorries, said: “Charging is getting more capital intensive and more competitive at the same time. That means two things decide who survives: the right sites and fast utilisation. If volumes do not ramp [up], payback stretches, assets get stranded and consolidation follows. That is just infrastructure market logic.”

The number of chargers installed in the UK has soared in recent years as companies raced to win market share. There were nearly 88,000 charge points across 45,000 UK locations at the end of 2025, according to the data company Zapmap.

Many charge point operators are making money, but others have installed points in anticipation of future demand, meaning they do not yet earn enough to cover costs, even if they are likely to as the number of electric cars on British roads rises rapidly.

«

This is either a) disastrous or b) just the natural part of the business cycle. If you compare the history of petrol stations, it took from 1919 to 1960 for filling stations to become established. At their height, there were about 40,000; then supermarkets came in, consolidation followed, and now there are about 8,500. Maybe we’re at the consolidation stage.
unique link to this extract

---

Under Trump, EPA’s enforcement of environmental laws collapses, report finds • Inside Climate News

Kiley Price and Marianne Lavelle:

»

Enforcement against polluters in the United States plunged in the first year of President Donald Trump’s second term, a far bigger drop than in the same period of his first term, according to a new report from a watchdog group. 

By analyzing a range of federal court and administrative data, the nonprofit Environmental Integrity Project found that civil lawsuits filed by the US Department of Justice in cases referred by the Environmental Protection Agency dropped to just 16 in the first 12 months after Trump’s inauguration on Jan. 20, 2025. That is 76% less than in the first year of the Biden administration. 

Trump’s first administration filed 86 such cases in its first year, which was in turn a drop from the Obama administration’s 127 four years earlier. 

“Our nation’s landmark environmental laws are meaningless when EPA does not enforce the rules,” Jen Duggan, executive director of the Environmental Integrity Project, said in a statement.

The findings echo two recent analyses from the nonprofits Public Employees for Environmental Responsibility and Earthjustice, which both documented dwindling environmental enforcement under Trump. 

From day one of Trump’s second term, the administration has pursued an aggressive deregulatory agenda, scaling back regulations and health safeguards across the federal government that protect water, air and other parts of the environment. This push to streamline industry activities has been particularly favorable for fossil fuel companies.

«

What’s the next step? Making lead compulsory in paint and petrol? (Though it’s probably removing fluoride from water, knowing the Madness of RFK.)
unique link to this extract

---

• Why do social networks drive us a little mad? • Why does angry content seem to dominate what we see? • How much of a role do algorithms play in affecting what we see and do online? • What can we do about it? • Did Facebook have any inkling of what was coming in Myanmar in 2016? Read Social Warming, my latest book, and find answers – and more.

---

Errata, corrigenda and ai no corrida: none notified