Supermarket shelves in the US could soon be empty of Italian-made pasta in response to Trump’s tariff hikes. Will frustrated buyers boil over? CC-licensed photo by Eli Brody 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.
There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.
A selection of 10 links for you. Unsalted. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
How catastrophic is it if the AI bubble bursts? An FAQ • The Ringer
Brian Phillips:
»
You’ve probably noticed that many people are saying we’re in an AI bubble. “OpenAI’s Sam Altman sees AI bubble forming,” CNBC reports. “Fears of an AI bubble are growing,” NBC News states. “AI Is the Bubble to Burst Them All,” Wired announces. “Mark Zuckerberg says a ‘collapse’ is ‘definitely a possibility,’” Fortune declares.
You’ve probably also noticed that almost everyone who uses the term “AI bubble” starts to sound very boring and confusing immediately afterward. This is because the people who bring up the AI bubble tend to be market analysts, and most market analysts would rather have potatoes growing inside their bodies than talk like a normal person for four seconds.
So what is an AI bubble? Are we in one? How scared should you be? What happens if the artificial intelligence industry blows a trillion-dollar hole in the non-artificial economy? What happens if it doesn’t? Let’s walk through this thing together.
1. Are we in an AI bubble?
Yes.2. How can you be so sure?
Because the gigantic numbers floating around the AI economy—deals worth trillions, data centers costing tens of billions, 75% of gains in the S&P 500 centered on AI stocks—are almost entirely driven by hype, and the hype has come unglued from reality. The hype around AI insists that it’s a world-transforming technology that will revolutionize every aspect of human society.The reality, which we’ll explore in more detail in a second, is that AI companies are burning through staggering amounts of money (and fossil fuels) with no clear path to profitability, that the companies themselves aren’t super clear about what their products are for, and that many of those products have failed to perform in the applications they’ve been assigned to (AI search engines return inaccurate information, AI teachers impair learning, AI therapists make mental health worse).
Worse yet for the industry, the biggest players are increasingly tied up in time-bomb financial deals that look disastrous for their futures in any but the rosiest of best-case scenarios.
«
The rest is a straightforward walkthrough of what we might expect. But what’s going to be left? Probably a much better electric grid. Which does have inherent value, like a railway or a lot of fibre optic cable.
unique link to this extract
Nvidia’s quiet move into quantum computing could reshape the next frontier of AI • The Motley Fool
Beegee Alop:
»
Quantum computing is still years away, but Nvidia just built the bridge that will bring it closer – a quiet integration of AI, GPUs, and patience that could shorten the wait for the next computing revolution.
Quantum computing is less a machine than a mission – a team of scouts sent to explore a landscape too complex to map by sight. Each scout sets out along a different path, testing what’s possible in parallel. Together, they can sense many routes at once – that’s the genius of the approach.
The challenge is keeping the team in contact. The radios crackle, the maps blur, and even a shift in weather can scatter their signals. These scouts – qubits – are astonishingly sensitive. They can explore multiple directions simultaneously, but the hardware carrying them is still too fragile for the conditions. A breath of heat or a tremor of noise can throw the expedition off course.
So instead of racing ahead, researchers spend most of their time stabilizing the mission: fixing equipment, recalibrating coordinates, and rerunning lost trails. The frontier remains open, but progress comes in slow, careful steps. That patience has defined the field – until now. And suddenly, the rhythm changed.
At its recent D.C. conference, Nvidia unveiled technology that could quicken that pace. Its new hybrid system – NVQLink and CUDA-Q – acts like a central command post for the scouts. It doesn’t ease the terrain, but it strengthens communication.
NVQLink connects quantum processors (the scouts) with today’s computing systems (the analysts) at microsecond speed – orders of magnitude faster than before. CUDA-Q, Nvidia’s open-source software layer, lets researchers choreograph that link – running AI models, quantum algorithms, and error-correction routines together as one system.
«
Nvidia is always looking for the thing after the current thing: when its big business was GPUs, it was looking at the potential in crypto and especially bitcoin; when that was hot, it was getting into the burgeoning generative AI business. And now that that’s hot..
One thing: this piece feels a bit chatbot-written to me. The author is said to be a “data engineer at The Motley Fool”. But this is the only article with that byline. Odd. (The LinkedIn profile – spoken languages Tagalog and English – is odd too.)
unique link to this extract
Apple TV execs dismiss introducing an ad tier or buying Warner Bros. Discovery • Ars Technica
Scharon Harding:
»
The heads of Apple TV have “no plans” to bring ads to the streaming service, balking, at least for now, at a strategy that has driven success for Apple’s streaming rivals.
In its November 2025 issue, British movie magazine Screen International asked Eddy Cue, senior vice president of Apple Services, if there are plans to launch an ad-based subscription tier for Apple TV. Cue responded:
»
Nothing at this time. … I don’t want to say no forever, but there are no plans. If we can stay aggressive with our pricing, it’s better for consumers not to get interrupted with ads.
«
The comments follow reports over the years suggesting that Apple has been seeking knowledge on how to build a streaming ads business. Most recently, The Telegraph reported that Apple TV executives met with the United Kingdom’s ratings body, Barb, to discuss what tracking ads on Apple TV would look like. In 2023, Apple hired advertising exec Lauren Fry as head of video and Apple News ad sales.
For Apple, “aggressive” pricing has meant three price hikes since Apple TV’s 2019 launch and a current monthly subscription fee of $13. For comparison, Netflix starts at $18 per month without ads, and Disney+ is $19/month without ads.
Introducing ads seems like a natural progression for Apple TV, not only because that’s what the competition is doing, but also because Apple TV reportedly doesn’t make money. Cue and the other Apple executives interviewed for Screen International’s article didn’t discuss revenue or profits or specify how many subscribers Apple TV has (Cue did say that Apple TV is “growing faster” and has more viewers with “more viewing hours in this past year than” ever before). In March, The Information, citing two anonymous people “with direct knowledge of the matter,” reported that Apple TV costs Apple $1bn per year. The publication’s sources claimed the service had about 45 million subscribers.
«
Last things first: that’s about $22 per subscriber, which isn’t intolerable money. And next: so there are places where ads would be intrusive or interrupting. And finally: that Screen International thing is right out of 1998: pages you can turn and everything.
unique link to this extract
The costs of instant translation • The Atlantic
Ross Benjamin:
»
To translate is not simply to transfer meaning but to attend to differences—of culture, time, thought, expression—that evade perfect alignment. I was reminded of this not long ago by a line from Bertolt Brecht, inscribed on a black stone pillar beside his statue outside the Berliner Ensemble theater: Die Veränderbarkeit der Welt besteht auf ihrer Widersprüchlichkeit. The line’s phrasing—“The changeability of the world rests on its contradictoriness”—is characteristically German, stringing together conceptually dense nouns. I pondered how I’d translate it, and each version I came up with tipped the balance differently: “The world’s capacity for change lies in its contradictory nature.” “The world can change because it is contradictory.” “The world’s contradictions make it possible to change it.” Such fine distinctions can’t simply be computed; they depend on being felt, weighed, chosen.
It’s in these very instances, when experience and intuition are essential, that language-prediction technology falls short. This is especially true in spoken communication. With its hesitations and half-understandings, it draws people into a shared project of meaning-making.
When I was living in Berlin, I realized that German offers specific opportunities for collaboration in conversation: Because the verb often comes last, you can set up the subject, object, modifiers, even the prefix, and leave the sentence hanging mid-word for your partner to complete. I’d begin, “Wenn man das System wieder ganz neu um- …” (“If one completely re- … the system”), and falter, my voice and face doing their best to convey what I was groping for. My partner might step in with umstellt (“rearranges”) or umbaut (“rebuilds”) or umdenkt (“rethinks”), and between us the sentence would find its footing. It wasn’t relayed fully formed from one of us to the other but took shape only through the exchange itself.
…While native English speakers might feel like they can afford not to engage with other languages, the wider world continues to pay outsize attention to Anglophone culture. I’ve seen this asymmetry in my field for a long time. Far more books are translated out of English than into it. The rise of instantaneous AI translation could make drifting into insularity even easier, allowing users to “understand” the world without ever leaving the comfort of their linguistic home.
Of course, this technology [of earbud translation] won’t just alter how we communicate; it also threatens to automate away an entire sector of skilled linguistic labor. Interpreters, translators, language teachers, subtitlers, and other specialists—people whose work relies on finely honed intellectual and creative abilities—now find our professions under mounting pressure from rapid technological change.
«
AI chatbots could help stop prisoner release errors, says justice minister • The Guardian
Rajeev Syal:
»
Release errors over the past fortnight have been seized upon by opposition MPs as evidence of the helplessness of ministers in the face of chaos within the criminal justice system.
David Lammy, the justice secretary, is expected to address parliament about the number of missing prisoners when MPs return on Tuesday.
It is understood that AI could be used to read and process paper documents; help staff cross-reference names to ensure that inmates are no longer hiding their past crimes behind aliases; merge different datasets; and calculate release dates and sentences.
At present, many of these jobs are being completed by inexperienced staff using calculators and reams of paper.
Responding to questions in the upper chamber on Monday, Lord Timpson said: “The number of releases per prison varies dramatically. In HMP Gartree, they average two releases a year, whereas … in Wandsworth it is 2,000.
“But that is why the digital team last week went into HMP Wandsworth to look at what are the opportunities for some quick fixes to embrace digital technology.
“We had the AI team that went in and, to give you a couple of examples, they think an AI chatbot would be really helpful, and also a cross-referencing for aliases, because we know some offenders have more than 20 aliases.”
He added: “We’ve given the team the green light to get on with that.”
«
“Inexperienced staff using calculators and reams of paper”. It says everything, doesn’t it? Nobody has been able to look at the system and say “why aren’t we using computers?” Too many layers and too little power for people to effect change.
unique link to this extract
AI-powered nimbyism could grind UK planning system to a halt, experts warn • The Guardian
Aisha Down and Robert Booth:
»
The government’s plan to use artificial intelligence to accelerate planning for new homes may be about to hit an unexpected roadblock: AI-powered nimbyism.
A new service called Objector is offering “policy-backed objections in minutes” to people who are upset about planning applications near their homes.
It uses generative AI to scan planning applications and check for grounds for objection, ranking these as “high”, “medium” or “low” impact. It then automatically creates objection letters, AI-written speeches to deliver to the planning committees, and even AI-generated videos to “influence councillors”.
Kent residents Hannah and Paul George designed the system after estimating they spent hundreds of hours attempting to navigate the planning process when they opposed plans to convert a building near their home into a mosque.
For £45-a-time, they are offering the tool to people who, like them, could not afford a specialist lawyer to help navigate labyrinthine planning laws. They said it would help “everyone have a voice, to level the playing field and make the whole process fairer”.
It is a modest enterprise but it is not alone. A similar service, Planningobjection.com, is promoting £99 AI-generated objection letters with the tagline “stop moaning and take action”.
Community campaigners have also encouraged supporters to use ChatGPT to craft objection letters on Facebook, claiming it is like having “a planning solicitor at your fingertips”.
«
UK government: hey everyone, let’s use AI, it’s the future!
NIMBYs: ooh yes look, lovely plan-stopping.
UK government: Not like that!
unique link to this extract
Multilingualism and extending healthspan • Ground Truths
Eric Topol:
»
What if learning a second language could provide you three more years of healthy aging or, alternatively, speaking only your mother’s tongue was linked to a loss of five years of healthspan? These are the implications of an important new study published on Monday. First, let’s review a bit of background.
Over the past two decades there have been many reports that multilingual individuals have improved cognition compared with their monolingual counterparts with better attention, task switching, working memory and potential protection from Alzheimer’s disease. It’s not just advantages for functionality. Brain structure studies in multilingual vs monolingual participants have shown increased gray matter density in key regions (caudate nucleus, left inferior parietal cortex) that are linked to executive function, such as this frequently cited 2004 paper [Figure in linked article]. This feature is known as neuroplasticity—the brain’s ability to change and reorganize its neural connections throughout life. Note how the gray matter density increase is maximum when proficiency in a second language occurs as a younger age.
Most of the studies of multilingualism’s benefits have been small, with confounders, and mainly focused on the brain and cognitive function. Now, the new report in Nature Aging provides a far more comprehensive picture.
«
Alors, c’est incredible! La deuxième langue donnent des années de vie. Mais seraient-t-ils bon?
unique link to this extract
Is AI a journalist or just a newsroom tool? • The New York Times
Benjamin Mullin and Katie Robertson:
»
Ryan Sabalow, a reporter for the newsroom CalMatters, noticed something peculiar when he began covering California lawmakers in 2023. Politicians would often give impassioned speeches against a bill, then refrain from voting entirely.
He began to wonder how often legislators were ducking tough votes — and how that influenced California’s laws.
Not long ago, those questions would send Mr. Sabalow scurrying to some dreary records room or scrolling through a spreadsheet. In the dawning age of generative artificial intelligence, all he had to do was ask a machine.
He and his team turned to an AI tool, Digital Democracy, which tracks every word uttered in California legislative sessions, every donation and every vote taken. It led to an article, and an Emmy-winning segment on CBS, that revealed that Democratic lawmakers had killed a popular fentanyl bill by not voting at all.
“I don’t think I could have done that without this database,” Mr. Sabalow said.
Artificial intelligence is sweeping through newsrooms, transforming the way journalists around the world gather and disseminate information. Traditional news organizations increasingly use tools from companies like OpenAI and Google to streamline work that used to take hours: sifting through reams of information, tracking down sources and suggesting headlines.
In some cases, including at Fortune and Business Insider, publications have explored using AI to write full articles, notifying readers they intend to use it for drafts.
Almost all of the news organizations have some guardrails in place to prevent errors, such as requiring a human to review anything that AI writes before it is published. But some embarrassing errors have appeared nonetheless, including from top publications such as Bloomberg, Business Insider and Wired.
«
Bit of a difference between Digital Democracy and the chatbotting work output at BI and Wired. One is a terrific resource – a true database – while the others just generate words. Lots of words. (Thanks Gregory B for the link. Gift article.)
unique link to this extract
Italian pasta is poised to disappear from American grocery shelves • WSJ via MSN
Margherita Stancati and Gavin Bade:
»
Your favourite Italian-origin fusilli and macaroni are poised to disappear from US supermarket shelves.
Italy’s biggest pasta exporters say import and antidumping duties totaling 107% on their pasta brands will make doing business in America too costly and are preparing to pull out of US stores as soon as January. The combined tariffs are among the steepest faced by any product targeted by the Trump administration.
“It’s an incredibly important market for us,” said Giuseppe Ferro, La Molisana’s chief executive, whose family-run pasta factory sits on the edge of the southern Italian town of Campobasso. “But no one has those kinds of margins,” he said, shaking his head as the sweet, nutty smell of freshly ground wheat berries permeated his factory.
The US Commerce Department has announced a 92% antidumping duty on pasta made in Italy by La Molisana and 12 other companies, which import the bulk of pasta from Italy to the US. That is on top of the Trump administration’s 15% tariff on imports from the European Union.
The Commerce Department acted after a long-running probe into pricing practices for the product that goes into everything from spaghetti Bolognese to mac and cheese. The severity of the decision has stunned one of Italy’s most iconic industries and has escalated into a diplomatic dispute between Washington and Rome, which is determined to combat the tariff.
«
Will Americans notice? Italian pasta exports to the US are worth $770m out of about $6.2bn – 11% or so. But there are plenty of American producers keen to step in; they’re the ones who have been complaining about dumping.
unique link to this extract
Churches respond to Nikalie Monroe after she exposed them • The Tab
Kieran Galpin:
»
In a 43-part saga that is taking over TikTok right now, influencer Nikalie Monroe exposed over 30 religious groups for ignoring the pleas of a fictional, desperate mother. Though there has been some light at the end of the tunnel, with some churches and mosques offering their assistance, the vast majority of groups refused and got called out as a result.
Some churches have even since responded, and it’s getting messier and messier by the day. Here’s the whole series, explained.
On Halloween, TikToker Nikalie Monroe started a series that would soon garner millions upon millions of views. She started with East Somerset Baptist Church in Kentucky with a simple premise: pretend to be a desperate mother seeking formula for her baby. Recording the phone calls for TikTok, Nikalie took on the role of a mother. She claimed to have run out of baby food and didn’t have the funds to purchase any. To really hammer home her point, she sometimes played an audio clip of a baby crying in the background. [She asked for formula, not money to buy formula, a very different thing – Overspill Ed.]
Though her explanations differed from video to video, some very much dependent on what the person on the phone said, more often than not the answer was no. Again, reasons differed greatly, but some included:
• We stopped doing that
• You don’t know anyone at the church
• You don’t attend the church
• We don’t have any
• We’re a church for old people
• Just straight up “no”
• You need to contact the local government
• You need to go to the store for thatThere have now been 43 parts in the series, with 10 offering to help and 33 rejecting her pleas. The first institutions to help were Islamic centres and mosques.
Following the social experiment, which doesn’t look to be cooling down any time soon, some churches and other religious properties have issued responses.
One pastor, who was from Baton Rouge, spoke about the Nikalie Monroe experiment in his sermon. He cited numerous times in the past that he’d helped people in need, but “rebuked” Nikalie’s phone call while calling it a “dirty deed.” He also said the TikToker was an “evil” witch who would be “dealt with swiftly” if she ever went to his church.
«
I’m.. not sure that I recall that last bit from the Bible.
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
The Overspill: when there's more that I want to say 