Plants do not wave warning flags when a hungry caterpillar shows up. Yet for years scientists wondered if roots guide underground networks that let one plant alert its neighbors before the first bite lands.
A modeling study released in January, 2025 suggests the picture is far less neighborly. Lead author Dr Thomas Scott of the University of Oxford argues that most plants would lose more than they gain by sounding a genuine alarm.
Why plants play tricks
Natural selection rewards thrift, so every calorie sunk into defense is a calorie that cannot go into growth or seeds. Because neighbors compete for the same sunlight and minerals, helping them makes little evolutionary sense.
The models draw on the idea of kin selection, the notion that genes “care” about relatives who share those genes. When the same genetic relatives also fight for the same patch of soil, the benefit of warning kin evaporates.
Simulations showed that honest signals kept disappearing from the virtual population. False alarms, however, sometimes stuck around long enough to handicap competitors by pushing them into needless chemical warfare.
“Our results indicate that it is more likely that plants will behave deceptively toward their neighbours, rather than altruistically, even when no herbivore is present,” said Dr Scott, adding that a plant might fake an attack draining rivals of resources in the process.
Fungus helps plants communicate
Much of the debate centers on the mycorrhizal fungi that lace through the soil, forming what ecologists nickname the “wood wide web.”
These threads trade phosphorus and nitrogen for plant‑made sugars and can physically link roots from different species.
Laboratory work with Solanum lycopersicum showed that when one tomato in a shared fungal network met a caterpillar, connected tomatoes cranked up four defense enzymes within hours. The reaction happened even though the “listener” plants never touched the insect.
Defensive cues may move as volatile organic compounds drifting through the hyphae or leaking from wounded roots.
Scott’s team argues that such cues are hard to suppress completely, creating an involuntary “tell” rather than an intentional SOS.
Model runs where the victim plant tried to hide the cue found that full suppression was rarely worth the metabolic fee. That fits the tomato result, where receivers benefitted even though senders paid the cost of damage.
If plants are unlikely whistle‑blowers, someone else must be moving the news. The models showed that the fungi themselves could profit by monitoring each host and relaying danger to all partners.
A healthy green plant pumps more sugar into the network than a chewed‑up one. By nudging the rest of the network into pre‑emptive defense, the fungus protects its own carbon income.
Hypotheses for information transfer in plant–fungal networks. Two plants are connected via a mycorrhizal network. One plant is under attack (e.g., by aphids), and this information may be transferred via the mycorrhizal network to the other plant, allowing it to up-regulate its defense mechanisms. There are three hypotheses regarding how this information is transferred: (blue) signaling by the attacked plant; (cyan) cues, which are potentially vulnerable to suppression by the attacked plant; and (orange) monitoring and signaling by the mycorrhizal network. Click image to enlarge. Credit: PNAS
“Maybe it is the fungal networks themselves that are sending the warning signals,” Dr Scott suggested. Co-author Professor Toby Kiers added that a neighbor may simply be “eavesdropping,” not altruistically sharing information.
Unlike deceptive plants, a cheating fungus would only hurt its paycheck. That steers evolutionary pressure toward honest fungal monitoring, at least until someone proves otherwise in field trials.
What this means for farms and forests
Agronomists dream of crops that switch on defenses before pests build an army. If fungi, not plants, drive advance warning, inoculating fields with the right spores might someday cut pesticide use.
The new study also tempers popular stories claiming that old trees “care” for their sapling offspring. Mutual benefit can still happen, but the math says it must be weighed against relentless competition for limited light.
Some breeders already select rootstocks that recruit efficient fungal partners. Future programs may screen those partners for their talent at early pest detection instead of simple nutrient delivery.
Forest managers experimenting with assisted migration should remember that uprooted seedlings leave their native fungal counselors behind. Re‑establishing key networks could be as vital as matching temperature or rainfall.
Rethinking plant behavior in evolution
These findings challenge the common tendency to describe plants using human social terms like friendship, cooperation, or mutual aid.
Such language can oversimplify complex ecological relationships shaped by evolutionary tradeoffs rather than intentional kindness.
Fungal networks help plants communicate and sometimes play “referee” to settle disputes. Click image to enlarge.
Plants don’t choose to help or harm. They respond to selective pressures based on survival and reproduction.
What might look like generosity from a distance could be a side effect of unavoidable signaling, or an outcome that benefits fungi more than the plants themselves.
Next questions for researchers
The models assume costs and benefits averaged over many generations. Field ecologists now need to tag real roots, real caterpillars, and real fungi to test whether the numbers hold up outside a computer.
One way is to pair isotope tracing with microelectrodes in split‑root systems, watching exactly who sends which molecule where.
Another is to compare fungal genomes for genes that sense host stress chemicals, a clue that selection favors monitoring hardware.
If dishonest plant signals do occur, greenhouse arenas seeded with mixed genotypes should reveal which seed families bluff the most.
Researchers may also seek fungal “referees” that penalize plants that persistently cry wolf. The soil is telling a subtler story than simple plant friendship.
The study is published in Proceedings of the National Academy of Sciences.
—–
Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.
Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.
Growing up, your mom probably told you to eat your fruits and veggies so you’d grow up to be big and strong. Now, a small study suggests another good reason to fuel up with produce: These nutrient-packed foods could help you snooze more soundly.
The observational research, which was published in Sleep Health, found that eating 5 cups of fruits and vegetables each day could result in longer, more uninterrupted sleep that same night, with fewer instances of waking up.
The findings show that better sleep doesn’t have to be complicated, says study coauthor Marie-Pierre St-Onge, PhD, a professor of nutritional medicine at the Columbia University Irving Medical Center in New York. “We can get better sleep with very simple things — with food.”
People Who Ate More Produce Slept More Soundly Through the Night
To tease out the effects of diet on sleep quality, researchers recruited 34 adults ages 20 to 49. About 4 in 5 participants were male, and all reported regularly sleeping seven to nine hours per night and eating three meals a day.
The participants contributed 201 days and nights of paired diet and sleep data, collected during their normal day-to-day routines. Each wore an activity monitor on the wrist to track sleep metrics and used an app to log everything they ate and drank for each 24-hour period.
The researchers then performed a nutrient analysis to see how participants’ daily diets affected their nighttime sleep.
They analyzed the effects of multiple variables, including:
Macronutrients: Carbohydrates (including fiber), protein, and fat
Micronutrients: Vitamins and minerals, including magnesium, B6, vitamin D, calcium, sodium, and zinc
Food groups: Meat, dairy, fruits and vegetables, nuts, and grains
The results showed that on days when participants ate more fruits, vegetables, and carbohydrates, they were more likely to get better sleep, which the study defined as having fewer instances of waking up during the night.
But not all carbs are equal when it comes to promoting sleep, says Dr. St-Onge. Simple carbohydrates like added sugars or the kind in white bread, for instance, will not help your sleep. “It’s the [complex] carbohydrates that also provide fiber and magnesium that are more helpful for sleep,” she says.
On the other end of the spectrum, on days when participants ate more red meat and processed meat, they had a higher likelihood of waking up multiple times in the night.
Sudha Tallavajhula, MD, a sleep neurologist with UTHealth Houston Neurosciences in Texas, who wasn’t involved in the study, says that the findings align with previous research showing that complex carbohydrates from fruits and vegetables help people fall asleep and stay asleep because they help the body absorb tryptophan.
The body uses this amino acid to make important neurochemicals, including melatonin, a hormone that tells the body it’s time to go to sleep. (Turkey, the beloved Thanksgiving staple, is rich in tryptophan, which explains why it sends many of us off to an afternoon nap.)
Fruits and Vegetables Provide Sleep-Inducing Fiber
Previous studies have found that a higher intake of fiber (a nondigestible carbohydrate that cannot be broken down into sugar) is connected to better sleep. A 2016 study, also coauthored by St-Onge, for instance, found that eating more fiber was tied to getting more slow-wave sleep, a deep, restorative stage of sleep that doesn’t involve dreaming.
Fruits and vegetables have large amounts of fiber, making them excellent choices for upping your sleep game.
One cup of raspberries, for example, has 8 grams of fiber. A single medium apple has 4.5 grams. The highest fiber-rich vegetables are green peas, broccoli, turnip greens, and Brussels sprouts.
The U.S. Department of Agriculture recommends 25 grams of fiber daily for women and 38 grams for men. Surveys suggest, though, that the average American adult gets only 16 grams of fiber per day.
Fruits and Veggies Also Have Good Amounts of Magnesium, Which Aids Sleep
St-Onge and her team found that magnesium was also tied to better sleep quality. “And where do we get magnesium?” St-Onge asks. “We get it from green leafy vegetables. We get it from complex carbohydrates, whole grains, and nuts and seeds,” she says.
Magnesium is an essential mineral well-known to promote sleep. “Magnesium stimulates the release of signals in the brain to relax and calm the body, and is also involved in the production of melatonin,” explains St-Onge.
Fruits and vegetables — you guessed it — are often rich in magnesium.
Tips on Incorporating More Fruits and Veggies Into Your Diet
Dr. Tallavajhula says that since the study was small (only 34 participants) and mostly involved men, it’s difficult to apply the results to everyone.
“But it’s an intriguing step in the direction of dietary advice,” she says.
Sharon Collison, RD, a clinical faculty member at the University of Delaware who specializes in medical nutrition therapy, says that the new study echoes what she sees in her practice: “I find that people who have erratic sleep patterns often have erratic food intake patterns. So improving one often helps the other one,” says Collison, who wasn’t involved in the study.
To get more fruits and vegetables into your daily diet, she recommends filling half of your plate or bowl with fruits and vegetables. Then fill another quarter with fiber-rich whole grains, such as oatmeal, whole-wheat bread, or brown rice.
If that feels like a dramatic increase for you, Collison says to get in the habit of eating a piece of fruit on your way home from work. This can also help reduce hunger when you are finally home — a time when many people struggle with snacking.
“I tell my patients that every meal and every snack needs to have a fruit and vegetable with it,” says Collison.
St-Onge adds that healthy eating should feel easy — and it shouldn’t break the bank either. You don’t have to buy everything fresh. “Even frozen is better than nothing at all,” she says.
Google’s Search Off The Record podcast discussed when a business should hire an SEO consultant and what metrics of success should look like. They also talked about a red flag to watch for when considering a search marketer.
Hire An SEO When It Becomes Time Consuming
Martin Splitt started the conversation off by asking at what point a business should hire an SEO:
“…I know people are hiring agencies and SEO experts. When is the point where you think an expert or an agency should come in? What’s the bits and pieces that are not as easy to do while I do my business that I should have an expert for?”
John replied that there is no one criteria or line to cross at which point a business should hire a consultant. He did however point out that there comes a certain point where doing SEO is time consuming and takes a business person away from the tasks that are directly related to running their business. That’s a point at which hiring an SEO consultant makes sense.
He said:
“Yeah, I don’t know if there’s a one-size-fits-all answer there because it’s a bit like asking, when should I get help for marketing, especially for a small business.
You do everything yourself. At some point, you’re like, ‘Oh, I really hate bookkeeping. I’m going to hire a bookkeeper.’ At that point where you’re like, ‘Well, I don’t appreciate doing all of this work or I don’t have time for it, but I know it has to be done.’ That’s probably the point where you say, ‘Well, okay, I will hire someone for this.’ “
SEO Should Have Measurable Results?
The next factor they discussed is the measurability of results. Over more than twenty-five years of working in SEO, one of the ways that low-quality SEOs have consistently measured their results is by the number of queries a client site is ranking for. Low-quality SEOs charge a monthly retainer and generate a report of all queries the site has ranked for in the previous months, including garbage nonsense queries.
A common metric SEOs use to gauge success is ranking positions and traffic. Those metrics are a little better, and most SEOs agree that they make sense as solid metrics.
But those metrics don’t capture the true success of SEO because those ranking positions could be for low-quality search queries that don’t result in the kind of traffic that converts to leads, sales, affiliate earnings or ad clicks.
Arguably, the most important metric any business should use to gauge the effect of what was done for SEO is how much more revenue is being generated. Keyword rankings and traffic are important metrics to measure, but the most important metric is ultimately the business goal.
Google’s John Mueller appears to agree, as he cites revenue and the business result as key measures of whether the SEO is working.
He explained:
“I think, for in SEO, it kind of makes sense when you realize there’s concrete value in working on SEO for your website, where there’s some business result that comes out of it where you can actually measurably say, ‘When I started doing SEO for my website, I made so much more money’ or whatever it is that goal is that you care about, and ‘I’m happy to invest a portion of that into hiring someone to do SEO.’
That’s one way I would look at it, where if you can measure in one way or another the effects of the SEO work, then it’s easier to say, ‘Well, I will invest this much into having someone else do that for me.’”
There is a bit of a problem with measuring the effects of SEO. The effects on sales or leads from organic SEO cannot always be directly attributed. People who are obsessed with data-driven decisions will be disappointed because it’s not always possible to directly attribute a lead from an organic search. For one thing, Google hides referral data from the search results. Unlike PPC, where you can track a lead from an ad click to the sale, you can’t do that with organic search.
So if you’re using increased sales or leads as a metric, you’ll have to be able to at least separate attributable paid search from earnings, then guesstimate the rest. Not everything can be data-driven.
Hire Someone With Experience
Another thing Mueller and Splitt recommended was to hire someone who has actual experience with SEO. There are many qualifying factors that can be added, including experience monetizing their own websites, ability to interpret HTML code (which is helpful for identifying technical reasons for ranking problems), endorsements and testimonials. A red flag, in my opinion, is hiring someone from a cold call.
John Mueller observed:
“Someone else, ideally, would be someone who has more experience doing SEO. Because, as a small business owner, you have like 500 hats to wear, and you probably can figure out a little bit about each of these things, but understanding all of the details, that’s sometimes challenging.”
Martin agreed:
“Okay. So there’s no one-size-fits-all answer for this one, but you have to find that spot for yourself whenever it makes sense. All right okay. Fair.”
Red Flag About Some SEOs
Up to this point, both Mueller and Splitt avoided cautioning about red flags to watch for when hiring an SEO. Here, they segued into the topic of what to avoid, advising caution about search marketers who guarantee results.
The reason to avoid these kinds of search marketers is that search rankings depend on a wide range of factors that are not under an SEO’s control. The most an SEO can do is align a site to best practices and promote the site. After that, there are external factors, such as competitors, that cannot be influenced. Most importantly, Google is a black box system: you can see what goes in, you can observe what comes out (the search results), but what happens in between is hidden. All search ranking factors, like external signals of trustworthiness, have an unclear influence on the search results.
Here’s what Mueller said:
“One of the things I would watch out for is, if an SEO makes any promises with regards to ranking or traffic from Search, that’s usually a red flag, because a lot of things around SEO you can’t promise ahead of time. And, if someone says, “I’m an expert. I promise you will rank first for these five words.” They can’t do that. They can’t manually go into Google’s systems and tweak the dials and change the rankings.”
Listen to Google’s Search Off The Record podcast here:
India has so far, refused to reconsider its decision to hold the Indus Water Treaty (IWT) in abeyance but the fact is that it cannot completely stop the flow of rivers into Pakistan, given the current infrastructure that it has, according to a report on Al Jazeera channel website.
But experts caution that even a small diversion or blockage could hurt Pakistan, if India were to manage to stop the flow of the Indus Basin rivers. They warn that any such move could set the stage for a full-fledged war between two countries.
In April, India said it was walking out of the IWT after gunmen killed 26 tourists in Indian Illegally Occupied Jammu Kashmir (IIOJK). A day later, Pakistan’s National Security Committee (NSC) rejected the “unilateral” move, warning that “any diversion of Pakistan’s water is to be treated as an act of war”.
The 85-page IWT brokered by the World Bank and signed in 1960 is different from most global water treaties that share water according to their total volume of flows. On the contrary, the IWT divides the rivers – three eastern rivers to India and three western rivers to Pakistan.
The treaty was a “hydraulic partition” that followed political partition, Majed Akhter, senior lecturer in geography at King’s College London told Al Jazeera. “It was needed to resolve issues of the operation of an integrated irrigation system in Punjab,” he added.
However, Akhter pointed out that water sharing between the neighbours is linked to their dispute over Kashmir. “Territorial control of Kashmir means control of the waters of the Indus, which is the main source of water for the heavily agrarian economies” of Pakistan and India, he added.
Rapid advances in artificial intelligence (AI) are reshaping economies and societies worldwide. While AI carries the potential for substantial productivity and welfare gains, it also presents significant risks, including job displacement, misinformation, cybersecurity, privacy breaches, and market concentration (Ben-Ishai et al. 2024, Comunale and Manera 2024). The dynamism of AI markets and the accessibility of cheaper and better-performing AI models to AI-adopting firms are critical conditions for the broad-based adoption of AI and long-term productivity gains (Acemoglu 2025, Filippucci et al. 2024, Aghion et al. 2024).
With the success of ChatGPT, many observers have feared that the excessive concentration of the technology within one or two leading US companies could reduce market dynamism (CMA 2024, Coeuré 2024, Cottier et al. 2023). This view was challenged by the release by the Chinese developer Deepseek’s of the R1 model in early 2025. This ‘Sputnik moment’ in AI (Acemoglu, 2025) stunned the world and highlighted several unexpected developments that have contributed to making AI more easily and more broadly accessible. It demonstrated that an almost unknown startup (The Guardian 2025) could train an AI model at the very top of AI capabilities at a fraction of the development costs of other leading models, publish it in ‘open-weight’,
and offer ten times cheaper access to users (Artificial Analysis 2025).
In this column, we provide novel empirical evidence to inform the debate on the state of competition in AI markets, primarily building on our paper (André et al. 2025) that collects novel data on generative AI prices and performance. The paper shows that during the past two years, the dominant position of digital incumbents in the AI supply chain (chips, models, distribution, and application) has neither curbed innovation nor prevented potential AI users from accessing better and cheaper AI models (Korinek and Vipra 2025, Hagiu and Wright 2025). The trends of better quality and more accessible AI create strong preconditions for adoption across various sectors (Bick et al. 2024), with significant implications for the economy and policymakers. While current trends give reasons for optimism about the diffusion of the technology, concerns remain about future concentration and market power, as happened with traditional digital markets.
Key trends in AI markets
Recent empirical evidence indicates a dynamic and competitive AI market landscape (André et al. 2025), benefiting companies that adopt AI technologies as an input in production, complementing and sometimes substituting for labour in the production of goods and services across several domains and a diverse set of cognitive tasks. The key trends in AI markets we document here column include:
1. A growing number of market players: The AI market has seen a surge in new entrants, including tech incumbents and specialised AI startups in model development (training of foundation models), provision (cloud infrastructure for inference), and applications (consumer-facing services).
2. A dynamic market environment: The pace of innovation in AI is extremely rapid, with continuous improvements in models that are more capable and reliable in an increasing number of tasks (including professional drafting, research assistance, software development, creation of image and videos, or real-time audio translations) and domains (science, law, finance, customer interactions, etc.). The current supply of AI offers a wide array of models, capabilities, and prices, allowing companies to choose their preferred quality and trade-off performance, costs, and privacy in their business operations.
3. Declining quality-adjusted prices: Advances in hardware and algorithms, as well as competitive pressure, have made AI models much more accessible. The best model in March 2023 (GPT-4) is now 1,000 times cheaper to access than two years ago. Our calculations show that the quality-adjusted AI price has decreased by 80% over the past two years. These trends include well-known text models (large language models) as well as specialised image or audio AI models.
New data and indicators to monitor AI markets
Our findings rely on an extensive data collection of AI foundation models available as-a-service on the market (accessible from the cloud to companies using AI in their production processes; see Bergemann et al. 2025 and Zhong 2025). In the period January 2023 to 2025, there were several indications of market dynamism in three segments of the AI value chain that we examined: AI model development, AI model provision from the cloud, and AI downstream applications.
First, the number of available AI foundation models has been rising exponentially (Figure 1), developed by an increasing number of companies and offering several interaction modalities (text, image, audio, or video). These results hold both for the number of cloud providers and downstream applications.
Figure 1 AI supply has been rising fast
Note: Developers refer to companies that train and optimise foundation AI models. The number of models refers to the number of active foundation models every month. Source: André et al. (2025).
The rapid pace of innovation has pushed out the economic frontier of AI
AI has improved and become cheaper at a similar pace as earlier general-purpose technologies (Filippucci et al, 2025). Using common industry benchmarks
to evaluate the performance of AI models and collecting prices of AI from cloud providers, we constructed an ‘AI economic frontier’, identifying the best models each month in terms of the price-performance ratio (Figure 2). Results suggest that in the last two years, this AI economic frontier has shifted continuously towards lower prices and higher quality. Moreover, the developers and models that reach the frontier have been changing, with five to six players alternating at the frontier (OpenAI, Google, Meta, DeepSeek, Anthropic, Mistral, etc.) and around ten others following closely. So far, leading positions in AI development are oligopolistic but highly contestable.
Figure 2 The AI economic frontier shows the continuous improvements of AI
Note: Performance is defined by a normalised weighted performance index on industry benchmarks. Each dot represents the model with the best available price-performance trade-off within Text-to-Text models. Source: André et al. (2025).
This variety of models at the frontier allows a broad range of demand to be served. Many users may not always need the best available models and would rather pay an order of magnitude less to access models that are ‘good enough’ and more easily specialised for specific tasks or preferences. In addition to the offer of closed models directly from the cloud, ‘open-weight’ models offer an option for cheaper (with no license fee), more transparent, and easily customisable models used outside of the public cloud environment.
The US has been pushing out the frontier, but other countries are in the race
Based on simulated market shares constructed using information on the supply of AI models (including price and quality), assumptions about demand for different AI market segments, and various scenarios involving switching costs and reputation, the US holds the leading position in AI development. In January 2025, the US market share for large language models is estimated to be between 86% (high switching cost scenario) and 59% (low switching cost scenario). China has been aggressively catching up since the second quarter of 2024, reaching simulated market shares of 5% (with high switching costs) and 36% (with low switching costs). The rest of the OECD countries combined (including Germany, France, the UK, and Canada) are further behind in AI text models (between 5% and 10%). However, they are in much better positions in specialised AI models (for example, image generation or speech generation), where their market share can jump above 50%, although in Europe and Canada, AI companies tend to be smaller, less well funded, and more specialised startups.
Figure 3 Simulated market shares in AI model development
Note: Simulated market share of AI foundation model revenues from AI as-a-Service per country of origin of the AI developing company under the baseline demand scenario and aggregating all modalities according to the formula in Annex C. In this scenario, 40% of AI demand is addressed to Text-to-Text models, 10% to Audio-to-Text and 50% to Text-to-Image. Source: André et al. (2025).
AI is getting better, cheaper, and more accessible for firms to adopt
The upward shift of the AI economic frontier depicted in Figure 2 is reflected in our quality-adjusted AI price index, which has declined by an average of 80% over the past two years (Figure 4), primarily due to continued quality increases offered at similar or somewhat lower prices. On average, during the period, 30% of models at the frontier are replaced every month with updated versions. These trends are consistent across all types of models, although the amplitude and timing vary.
In sum, this evidence suggests that the gains from innovation in AI development have been shared with AI users in terms of lower quality-adjusted prices. AI users have also benefited from greater access to AI models via a widespread offer accessible through several cloud providers (for business use) and an increasing number of AI-powered consumer services (consumer-facing applications).
Figure 4 Quality-adjusted AI prices have fallen rapidly
Note: The index for each modality is a weighted sum of the index for each model segment. Each model segment is represented by its respective model at the AI Economic Frontier. Source: André et al. (2025).
Why these trends matter
Dynamic AI markets are a necessary condition for the diffusion of AI across the economy, facilitating widespread AI adoption in various sectors and serving as a central determinant of long-term productivity gains from AI (Acemoglu 2025, Filippucci et al. 2024, Aghion et al. 2024). Our evidence so far suggests that the supply of AI has been more open than initially expected in various segments of the AI value chain, driving innovation and contributing to favourable conditions for broad AI adoption: lower prices, better quality, and broader accessibility.
Nonetheless, several uncertainties and risks persist about the future dynamism of AI markets. For instance, the capacity of incumbents to leverage existing compute infrastructure and user base in adjacent markets is high. Furthermore, the high concentration of the necessary inputs for AI development – data, compute, and talent – creates additional risks for long-term competition, which warrant continued research on the dynamism of AI markets.
References
Acemoglu, D (2025), “A Sputnik moment for AI?”, Project Syndicate, 4 February.
Comunale, M and A Manera (2024), “The Economic Impacts and the Regulation of AI: A Review of the Academic Literature and Policy Actions”, IMF Working Papers 2024(065).
Acemoglu, D (2025), “The simple macroeconomics of AI”, Economic Policy 40(121): 13-58.
Aghion, P and S Bunel (2024), “AI and Growth: Where do we stand”, unpublished manuscript.
André, C, M Betin P Gal and P Peltier (2025), “Developments in Artificial Intelligence markets: New indicators based on model characteristics, prices and providers”, OECD Economics Department Working Paper.
Artificial Analysis (2025), Q1 2025 state of AI report: Analysis of the AI landscape and the key trends shaping AI.
Ben-Ishai, G, J Dean, J Manyika, R Porat, H Varian and K Walker (2024), “AI and the Opportunity for Shared Prosperity: Lessons from the History of Technology and the Economy”, arXiv preprint arXiv:2401.09718.
Bergemann, D, A Bonatti and A Smolin (2025), “The Economics of Large Language Models: Token Allocation, Fine-Tuning, and Optimal Pricing”, arXiv preprint arXiv:2502.07736.
Bick, A, A Blandinand D J Deming (2024), “The rapid adoption of generative AI”, NBER Working Paper No. w32966.
CMA – Competition and Market Authority (2024), AI foundation models technical update report.
Coeuré, B (2024), “Comments on “The simple macroeconomics of AI” by Daron Acemoglu”.
Cottier, B, T Besiroglu and D Owen (2023), “Who is leading in AI? An analysis of industry AI research”, arXiv preprint arXiv:2312.00043.
Filippucci, F, P Gal and M Schief (2024), “Miracle or Myth? Assessing the macroeconomic productivity gains from Artificial Intelligence”, OECD Artificial Intelligence Papers, No. 29.
Filippucci, F, P Gal, K Laengle and M Schief (2025), “Macroeconomic productivity gains from Artificial Intelligence in G7 economies”, OECD Artificial Intelligence Papers No. 41
Hagiu, A and J Wright (2025), “Artificial intelligence and competition policy”, International Journal of Industrial Organization, 103134.
Korinek, A and J Vipra (2025), “Concentrating intelligence: Scaling and market structure in artificial intelligence”, Economic Policy 40(121): 225-256.
OECD (2024), “Artificial intelligence, data and competition”, OECD Artificial Intelligence Papers No. 18.
The Guardian (2025), “Who is behind Deepseek and how did it achieve its AI Sputnik moment?”, 28 January.
Zhong, H (2025), “Optimal Integration: Human, Machine, and Generative AI”, CEPR Discussion Paper No. 20330.
I was delighted to read your column, “How indigenous art became in demand” (The Art Market, Life & Arts, June 14).
For decades, social enterprise academics and practitioners have championed fairness, equality and diversity by advocating for better market access for farmers in the global south. We have long argued that trade — not aid — is the sustainable path to shared prosperity and development.
From Fair Trade bananas to Fair Trade indigenous art, we have campaigned to correct market, governmental and societal failures that deprive millions of access to basic goods, services and fair economic participation. This column highlights one such failure — the exclusion of indigenous art from global markets by the dominant institutions of the art world.
Thankfully, the tide is turning.
The growing movement to decolonise the arts is beginning to open doors — enriching the global cultural landscape and bringing long-overdue recognition to indigenous artists. This shift benefits not only the artists and their communities but also collectors, galleries and audiences who seek new and diverse perspectives.
The market for indigenous art is still emerging, and its full potential is yet to be realised. In the future, those who fail to value these rich artistic traditions may find themselves lacking in the cultural capital that increasingly shapes opportunity and influence.
Colin Jost is sharing how he really feels about those viral red carpet kisses between Jurassic World Rebirth co-stars Jonathan Bailey and Scarlett Johansson, who is also his wife.
In a recent interview with Entertainment Tonight, the reporter and Jost discussed the press surrounding the movie and things he “didn’t expect” to become a story, like the co-stars’ public display of affection.
The SNL star playfully said, “I guess in Jurassic Park terms, the attack always comes from the raptor you never thought [of.] Of all the threats out there, I wasn’t thinking it was Jonathan.”
“People really blow it out of proportion. When someone, like, kisses their friend hello. That’s pretty nuts,” Jost said. “Jonathan’s an out gay man. It didn’t seem like the biggest threat. Jonathan and I were like, ‘I guess we have to kiss now? Is that what happens?’ Close the loop!”
Colin Jost reacts to Jonathan Bailey and Scarlett Johansson kissing in the only way he knows how – ‘Jurassic Park’ jokes. 🤣 pic.twitter.com/6EGenbNz1y
The Black Widow actress and Wicked actor made headlines when they were photographed smooching on the red carpet at the film’s London and New York City premieres last month.
Following the appearances, Johansson stopped by the Today show, where she agreed with co-anchor Craig Melvin that their kisses were platonic and friendly. “He’s a lovable guy, what can I say?,” she noted when speaking about Bailey. “I don’t know, we’re friendly people.”
She was also asked about the internet reactions to the kisses. “Nothing surprises me, you know what I mean?” she responded. “Nothing surprises me these days. But yeah, I’ve got a lot of love to give, what can I say?”
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
UK patients are less likely to receive advanced treatments for lung cancer than counterparts in similar countries, potentially contributing to Britons’ lower survival rates, according to the first study of its kind.
The research found that in parts of Canada, just over half of patients in the later stages of the disease received chemotherapy, compared to 45 per cent in Norway but only 43 per cent across the UK. In Northern Ireland, the proportion stood at just 23 per cent.
Radiotherapy use was also lower in the UK compared to the average across all stages of lung cancer. The proportion of patients receiving the treatment ranged from 32 per cent in England to 50 per cent in areas of Canada.
Michelle Mitchell, chief executive of Cancer Research UK, acknowledged that more research was needed to understand what had led to the treatment gaps.
But, she noted, the struggling national health service could be a factor, pointing to “the pressures that the NHS is under, with lots of challenges around staff and resources”.
The study, led by the International Cancer Benchmarking Partnership (ICBP) of international clinicians, policymakers and data experts, compared the use of chemotherapy and radiotherapy in a cohort of 280,000 patients from Britain, Australia, Norway and Canada.
People in the UK also waited longer for treatment compared with international counterparts. Fewer than 15 per cent of patients received chemotherapy within 30 days of diagnosis, compared to more than 22 per cent in Norway.
Improvements in cancer survival across the UK have fallen sharply in recent years, according to experts, with survival rates in the country lagging peer nations for certain categories of cancer.
Cancer Research UK, which managed and part-funded the work, highlighted recent ICBP research that found five-year net survival for lung cancer in the UK stood at just 14.8 per cent.
This compared with 20.4 per cent in Norway, 21.3 per cent in Australia and 21.7 per cent in Canada.
The study noted that “differences in non-surgical treatment use are plausibly associated with international variation in lung cancer survival”.
Another issue might be late diagnosis, suggested Mitchell. Patients might already be in such poor health when their condition was spotted that it was “not tolerable or medically advised” to undergo the rigours of more advanced treatment.
Some patients might also face a long journey to reach a specialist centre, creating another barrier to treatment.
Mitchell called on ministers to commit to rolling out lung cancer screening to all 55-74 year olds who have smoked across England by 2029, followed by the three other constituent UK nations.
The NHS 10-year plan published last week committed to a rollout but failed to give a timeframe, despite having previously said it would happen by 2030. The government has promised to produce a national cancer plan later this year.
“We need to use this opportunity of the national cancer plan to transform cancer outcomes across the board so we’re world-leading, rather than lagging behind comparative countries,” Mitchell added.
The data, the most recent available that spans multiple countries, covers the period from 2012 to 2017. But Mitchell said she did not think the overall picture had changed in recent years.
The Department of Health and Social Care said it was introducing measures to bring cancer care closer to home and to catch cancer earlier and quicker.
The reforms would “will see around 120,000 more people referred for urgent cancer checks, get a diagnosis within four weeks and start treatment within two months”, it said.
In this week’s story, a recent grad enjoys her last days of freedom before starting a new job: 24, single, Europe.
DAY ONE
10 a.m. I’m in a museum in Copenhagen, and it’s pouring outside. I’m with a platonic friend who I met on a train to Stockholm. Soon I have to leave for the train station. It’s weird to say good-bye because I’ll probably never see him again.
1 p.m. I arrive at the train station to take a train to Berlin. I’ve been solo traveling around Europe for about two weeks now, embarking on a grand trip after graduating from my master’s program.
4 p.m. It’s cozy to be in the train while it’s still pouring rain outside. I read my giant copy of 1Q84, by Haruki Murakami — the one book I took with me on my trip with the intention of it lasting my entire monthlong expedition.
9 p.m. I’m walking to my hostel in Berlin, and the sunset over the river is mesmerizing. I immediately feel that this place is alive.
10 p.m. When I get to my room in the hostel, I introduce myself to the girl sleeping in the bunk above mine. A is from Brazil but studying abroad in Portugal. The table in the entryway of the room is littered with empty vodka bottles, food crumbs, and rhinestones.
10:30 p.m. Some girls from England are sharing our room, too. They’re getting ready to go clubbing. One girl strips down, putting on a lace thong and fishnets. She’s worried her outfit isn’t good enough to get into the Kit Kat Club. Her friend echoes these concerns — they need to look slutty but tasteful.
11 p.m. A and I go to the hostel bar in the hopes of starting an interesting evening. As we get to know each other, we catch the attention of some English men. They’re here for some football event. I am chastised for knowing nothing about it.
12 a.m. We follow the English guys to the bar next door, where they buy us espresso martinis. One of them, T, continues to tease me in a flirty way: You Americans say that, you Americans do this. I’ll make fun of myself for free drinks — actually, I’ll make fun of myself for free. I like T’s attention, but I feel him losing interest. Another man, M, is trying to get cozy, but I’m not feeling it.
2 a.m. A decides to go to bed. She’s more sensible than me. Now I am the lone girl in a group of rowdy guys. We walk through the city looking for a lively party. Although I don’t see myself actually wanting to hook up with any of them, or even see them ever again, their commitment to the chase fills me with glee.
3 a.m. We’re at a sports bar with sticky tables. The night can go nowhere good from here, so I decide to walk home. I tell them I don’t need an escort, but M insists on walking me back to the hostel. T hangs back. I am disappointed by this turn of events. As we reach the hostel, M tries to kiss me, but I duck away, thanking him for the drinks. I escape into the hostel lobby, where two guys watched the whole exchange. They are amused at my lack of interest, and one of them, Y, invites me to the Kit Kat Club tomorrow night. I accept.
DAY TWO
8 a.m. I didn’t sleep much, and I can feel it.
9 a.m. The English girls return from the club. One of them starts loudly talking about how she had sex on the dance floor. I try to ignore them and tell A about the rest of my night and the random invitation to the Kit Kat Club. One of the English girls looks me up and down and scoffs, “Do you even have an outfit?” I laugh nervously. She tells me to head to a lingerie store.
11 a.m. I take myself to a museum to see an exhibition by Caspar David Friedrich. His art moves me. I graduated with a degree in environmental engineering about a month ago. When I return home, I start a full-time job in engineering. I try to savor my freedom while I can.
1 p.m. I meet a family friend, R, for lunch. R lives here now (she moved for a man; it’s complicated), and I really enjoy her company.
3 p.m. Together, we find one of the few lingerie stores nearby that’s open on Sunday. We’re the only customers. I look at piece after piece, feeling, well … broke. I try on a couple of outfits that are on sale and hate the way I look. I’m not going to get into the club. I buy the cheapest option, a red pleather zip-up minidress and crotchless fishnet tights.
9 p.m. R and I eat some delicious Indian food, then I run back to the hostel to get ready.
11:30 p.m. The Uber drops me off in a dark and empty part of the city. I walk around looking for the club, surprised there isn’t a line out the door. A man sitting on the street corner offers me crack. I decline. I walk up to the door where I think the club is located. I’m nervous but excited. I walk in, hand over my phone — no phones are allowed in the club — and walk over to the dance floor. I see Y, the guy I met the night before at my hostel, dressed only in a leather thong, and give him a hug. I like that everyone is given a chance to be someone else here.
12 a.m. While I’m waiting to order, I’m approached by a tall man with curly blond hair and thick glasses. He’s wearing a mesh bodysuit decorated with a geometric pattern of dicks and mesh booty shorts. His arms are covered in wacky bird tattoos. He buys us both Moscow mules and gives the bartender a generous tip. We start talking and are immediately comfortable with one another. I compliment his tattoos, touching his arm, and ask him to explain them all to me. He is from Montréal, here on holiday, and I am taken with his French accent. He wants to check out the rest of the club and suggests we walk upstairs.
12:15 a.m.“Do you like making out?” he asks me. Yes, of course, I reply. He’s muscular and a really good kisser. He’s pulling down my crotchless stockings, and I’m thinking about how many people can see my ass right now.
12:45 a.m. I want to dance, so I drag him to the dance floor and I grind on him to the techno beat.
1:30 a.m. We walk upstairs to make out. He keeps lifting me off the love seat and setting me back down again, which is really hot. He starts fingering me, and I think about the people that might be watching us. He asks if I want to go back to his hotel, and I say yes.
2:30 a.m. The streets are empty, and I feel safe with him. Conveniently, his hotel is a five-minute walk from the club, and he’s also traveling alone. I ignore any weirdness about the situation and lean into the moment.
2:45 a.m. The hotel turns out to be very fancy. When we get to his room, he asks to take a shower with me. He shampoos my hair. His room is messy and he has neat rows of pill bottles on the counter. He explains that his antidepressants might make it difficult for him to get aroused. I try to calm his worries. It’s exciting just to be with him. But then I’m able to arouse him and … “You’re big,” I tell him. He does a good job of turning me on, and the pain of penetration, due to his size, is pleasurable.
3:30 a.m. As I’m falling asleep in his arms, I plot ways in my head to figure out his name the next morning. I’ve totally forgotten it.
DAY THREE
9 a.m. I wake up to a light-filled room. He offers me clothes so I don’t have to go back to the hostel in last night’s outfit. I drown in his Uniqlo sweatpants and T-shirt. They’re cozy. With a bit of hesitation and shyness, we discuss spending the day together, and it becomes clear we both want to. I go back to the hostel to change and shower. I give him my phone with the contact open so he can type in his own contact information. When he hands it back to me, I learn his name — I’ll call him B.
11 a.m. I meet B at an avocado-toast restaurant. As I show up in a green knotted headband, regular T-shirt, and shorts, I think about how the image of me from last night — crotchless tights and a red pleather minidress — is not at all what I usually look like, and B has no idea what I’m like in real life. We start to get to know each other over breakfast. He’s quite affectionate and insists on paying for my meal. We decide to go to the Jewish History Museum together.
1 p.m. B and I explore the museum. He takes time to read everything, looking at objects closely. It’s a heavy experience.
3 p.m. We go back to the hotel to “take a nap,” a.k.a. have sex.
5 p.m. We take the train to an outdoor graffiti gallery recommended by my friend, then we stop for pizza. We drink wine, eat, and talk. Then we see a biergarten, grab beers, and take them down to the river’s edge. I snap a photo of him holding a cigarette; he’s beautiful. We keep learning about each other, opening up. He tells funny stories about his friends in Montréal. My affection for him grows.
10 p.m. We meet R at an outdoor movie. When the movie is over, we go grab a beer at an outdoor hut. I can tell that R is incredibly amused by the situation. We go back to B’s hotel room to go to bed.
DAY FOUR
10 a.m. We eat breakfast in the hotel lobby. I’m going to Prague today, with a plan to stay for three nights before moving on to the next city. I ask B to come with me. He has a few hours to figure it out while I go to the hostel to pack up. He says he’s going to look into train tickets.
12 p.m. I pack up the rest of my belongings at the hostel. B texts that he’ll meet me at the train station! We barely make the train. But after we board, we hold hands and look out the window and listen to music together.
5 p.m. We check into the private room B booked for us in a hostel in Prague. I’m nervous because this is a bit of a commitment — we’re staying together for the next three days. The room is luxurious after weeks of hostel bunks.
6 p.m. We are both tired and hungry. We try to find a restaurant, and he’s annoyed that there are so many places I can’t eat at because they don’t have vegetarian options. But we find a place where I eat a delicious salad, and the conversation is mellow but engaging.
8 p.m. We walk the river together and sit on the rocks. He rolls a couple of joints. We smoke and I barely get high — the weed is really bad.
10 p.m. We buy some Ben and Jerry’s on the way to the hostel and watch Netflix in bed.
DAY FIVE
10:30 a.m. We find a breakfast place that looks good. I eat a frittata.
12 p.m. We take a streetcar to the base of a hill where there’s a walk that was recommended for good views. We have someone take our photo. We look cute together, and I feel affectionate toward him.
3 p.m. We walk through Old Town. B picks up a canned cocktail from a vendor on the street. We pass a dispensary, and he wants edibles. He takes a few as soon as he buys them.
4 p.m. We stop for wine and olives. I feel a deep exhaustion taking root within me.
5 p.m. We go for more wine and appetizers near our hostel. B keeps insisting on paying, which is nice.
7 p.m. B and I go back to the hostel. He randomly meets a fellow Canadian at the bar, and we start talking. More people join us. It seems they are all solo traveling. I immediately like one of the women, J. She’s spunky, smoking a cigarette, and I can tell she’s fearless. She’s on a three-month Europe trip before heading to Asia for another three months. I’m jealous and have no idea how she can pull off this lifestyle. We all decide to go out together and find a jazz club (the thing to do in Prague). Kinda feels strange that B and I are acting like such a couple. To be viewed as a couple, externally, with someone I just met is a weird experience I’ve never had before.
11 p.m. After jazz, we grab beers at a crowded sports bar. B is incredibly drunk. He leaves to go to the bathroom, and I suspect he’s throwing up. The night is going nowhere good. Soon after, we go back to the hostel.
DAY SIX
12 p.m. We sleep late.
1 p.m. We make our way to a botanical garden nearby, but when we pass a winery, B says he wants some wine. I agree, but so much alcohol over the course of the week is deepening my feeling of exhaustion. Today is our last day together — I had plans to travel to Vienna next, and B intended to go back to Berlin. I try to fake a happiness with him that I don’t quite feel anymore.
2 p.m. We stroll a garden and sit on a bench. I get a picture of us.
4 p.m. B and I are back in the room. Both tired. He wants to smoke weed, so we do. Afterwards, I feel nauseous and anxious. He can tell I’m upset and doesn’t know what to do. I lie down and try to rally to enjoy our last evening together.
8 p.m. We sit inside a vegan restaurant. B immediately orders a negroni. I truly feel like our fever dream is coming to a conclusion. Two souls from two different countries who alleviated each other’s loneliness for a brief moment in time.
10 p.m. B and I have sex one last time. It’s not bad, but it doesn’t make me feel anything.
DAY SEVEN
10 a.m. B and I eat our last breakfast together at a little café we’ve been eyeing near the hostel. The moment is bittersweet.
12 p.m. We arrive at the train station together. He’s leaving first, so I walk him to the platform. We hug and he says he’ll see me in Montréal, which we both know won’t happen, but in this moment, we can’t entertain the possibility that we’ll never see each other again.
12:30 p.m. My train is full and I must wait until the next one in a few hours. It’s weird to be alone now. I go to a coffee shop nearby and touch base with all my friends who were wondering what happened with the man I met in the club.
5 p.m. I’m finally on the train. Exhausted. I listen to music and gaze out the window. B texts me a French song that I end up loving and listen to on repeat.
9 p.m. My hostel in Vienna is a bit outside the city, and it’s quiet when I arrive. I feel sweaty and gross. I sit eating a protein bar in the common area and looking at the people around me. Who will I meet next, and what adventure will they take me on?
Want to submit a sex diary? Email sexdiaries@nymag.com and tell us a little about yourself (and read our submission terms here.)
CNBC’s Jim Cramer on Friday told investors what to follow next week as earnings season kicks off, highlighting reports from JPMorgan, Netflix, Goldman Sachs and PepsiCo.
“Once we process the new tariffs, we’ve got a ton of earnings reports coming next week, so you better keep your eyes open,” he said.
Tuesday brings earnings from financial giants JPMorgan, Wells Fargo, Citigroup and BlackRock, and Cramer said he’ll be waiting to hear whether there has been any slowdown in spending or pick up in loan losses. While he said JPMorgan is “the star of the show,” he also cares about Wells Fargo, which is no longer subject to a punitive asset cap. Cramer predicted Citigroup’s report would be well-received, but he said BlackRock might tell “the most exciting story.” The Labor Department will release the consumer price index report on Tuesday, Cramer added, an important metric for the Federal Reserve when it makes decisions about interest rates.
On Wednesday, Goldman Sachs and Morgan Stanley are set to report, and Cramer said he’s optimistic both outfits will post strong quarters as mergers and acquisitions heat up. He said there could be “another round of semi buying” if semiconductor capital equipment company ASML releases a solid report. Bank of America and Johnson & Johnson will report on Wednesday as well. Cramer said Bank of America has put up consistently good earnings and suggested the stock was cheap because Berkshire Hathaway has been selling shares. Cramer said the pharmaceutical giant still has litigation hanging over its head.
Retail sales figures will come out Thursday, and Cramer said he’s worried about a slowdown as political chaos affects consumers. Abbott Laboratories, PepsiCo and Netflix are set to report Thursday. Cramer said Abbott’s quarter tends to be “misinterpreted in a negative way,” saying the healthcare name is one of his favorite companies. Cramer called PepsiCo “too cheap relative to its growth rate,” and he noted different factors that could be weighing on the stock, like the rise of GLP-1 weight loss drugs and scrutiny on junk food from Health and Human Services Secretary Robert F. Kennedy Jr. Cramer said he bets Netflix will report a great quarter, but he said the bar for the streaming giant is high.
On Friday, American Express, 3M and Charles Schwab will report earnings. According to Cramer, American Express tends to sell off even when the report is good. The industrial sector has been doing well lately, and Cramer said 3M might report one of the best quarters of the group. He was also optimistic about Charles Schwab, but said “the short-sellers like to come out and color the opening of trading when Schwab opens,” advising investors to be careful before they buy.
Jim Cramer’s Guide to Investing
Sign up now for the CNBC Investing Club to follow Jim Cramer’s every move in the market.
Disclaimer The CNBC Investing Club Charitable Trust owns shares of Wells Fargo, BlackRock, Goldman Sachs and Abbott Laboratories.
