A proprietary survey shows AI is fueling automation that’s changing the trajectory of early-stage pharma research, Axios’ Adriel Bettelheim writes.
Why it matters:Â The TD Cowen survey of 80 biopharma leaders and insiders finds AI is reducing drug developers’ preclinical costs and timelines by as much as 70%.
That’s fueling demand for cutting-edge software, sequencing tools and computer models that will help churn out more experimental treatments in the next five years
. The big picture: AI can’t replace scientific intuition. But the way it can streamline the grueling R&D before human testing begins is making computer screens as important a feature of drug design as traditional laboratories.
Brendan Smith, director of life sciences equity research at TD Cowen, says the hope is to generate large amounts of data to train the AI models and increase the odds of clinical success.
One of China’s leading AI companies may have leapfrogged the industry’s frontier labs.Â
Moonshot AI recently released Kimi K3, the latest iteration of its flagship model family, and one benchmark claims it has outflanked Anthropic, OpenAI, Google, and others.
Kimi K3 features 2.8 trillion parameters, a one-million-token context window, and a multimodal architecture that comprehends text, images, and video within a single model. The full open weights will be available on July 27, and for now the model is usable via Moonshot’s API.Â
But Moonshot’s release may represent a change in the tides:Â
According to Arena.ai, a benchmarking and evaluation platform for frontier models, Kimi K3 has taken the top spot in 6 of its 7 frontend domains, including brand and marketing, reference-based design, data and analytics, consumer product, simulations, and content creation tools. The model surpassed the performance of the latest models from Anthropic, OpenAI, xAI, Meta and China’s Z.ai in these categories. Kimi K3 only ranked second in the gaming domain, slotting just behind Anthropic’s Claude Fable 5. The score represents a 17-place jump from the company’s previous model generation, Kimi-K2.6, according to Arena.Â
The results have sent shockwaves through the industry in recent days: Gavin Baker, managing partner and CIO of Atreides Management, said in a post on X that the release marks an “inflection point” for AI that could spell trouble for Anthropic and OpenAI by breaking up their dominance. Cisco CEO Jeetu Patel posted that improving models like these could lead to more competitive industry dynamics by creating a gross margin drop for frontier models. David Sacks, co-chair of President Trump’s Council of Advisors on Science and Technology, said in a post that the release is “concerning” in light of calls to regulate AI in the US and that “this is how you lose the AI race.” Â
It’s important to note that Arena’s benchmark results differ from those that Moonshot itself released. In its post announcing Kimi K3, the company said that the model sat just below Fable 5 in benchmarks for coding, agents and frontier software engineering, and only outranked Anthropic’s model in benchmarks for codebase cleaning and long-horizon engineering.Â
It’s also important to note that Moonshot is one of three Chinese labs, along with DeepSeek and MiniMax, accused of “industrial-scale” model distillation campaigns by Anthropic in February, taking part in more than 16 million exchanges with Claude through 24,000 fraudulent accounts to acquire and use its capabilities.Â
“Moonshot is marketing Kimi K3 with new architecture work unique to its lab, but that could be used in conjunction with distillation techniques as part of training a model,” Brian Jackson, principal research director at Info-Tech Research Group, told The Deep View.
Though major industry voices are sounding the alarm on Moonshot’s advancements, the natural tendency in such a fast-paced industry is to get excited by every shiny new toy that hits the market. However, Kimi K3 may not be evidence enough that China is pulling ahead. Though the nature of open-source technology broadly has always been to take the work of another and improve upon it, given the model distillation accusations by closed-source, proprietary labs, we need to question whether or not models from these major Chinese labs are powerful on their own accord, or simply powerful because they are piggybacking off of the research of the leading frontier AI labs.Â
Historian Bill Federer: “By 2030, there will be a majority Muslim population in Europe, and they will just flat out vote for Sharia law. People forget Egypt was completely Christian for 6 centuries. It’s not anymore. All of North Africa was completely Christian for 6 centuries. It is not anymore. Constantinople was the largest Christian city in the world, and the largest Christian church in the world for hundreds of years was the Hagia Sophia. And it was turned into a mosque. They want to do the same thing with the Vatican.” Pay attention to what he is saying!
Denmark, particularly in Copenhagen, has installed floating platforms/islands planted with wildflowers and native vegetation in its harbors and canals. These create safe urban habitats and sanctuaries for birds, bees, pollinators, insects, and even some aquatic life.
From colonial symbol to sanctuary: Paris’s Grand Mosque turns 100
Interview
France
The Grand Mosque of Paris, built in 1926 in the Latin Quarter as a tribute to Muslim soldiers who died for France during World War I, marks its 100th anniversary this year. French historian Benjamin Stora traces its little-known role as a geopolitical bridge â not only between France and Algeria, but also between Muslims and Jews.
Tucked away in the heart of the Latin Quarter neighbourhood, the Grand Mosque of Paris is one of the cityâs essential landmarks. Inaugurated on July 15, 1926 by former French president Gaston Doumergue and Moulay Youssef, Moroccoâs sultan at the time, the mosque was built as a tribute to the more than 100,000 Muslim soldiers who fought and died for France during World War I.
The Mosque‘s Hispano-Moorish architecture boasts a 33-metre-high minaret, courtyards and interior gardens inspired by Arab-Andalusian palaces and porticoes embellished with tiles hand-carved by Moroccan craftsmen in 1922. A jewel of Islamic architecture in France, the Grand Mosque of Paris has been listed as a historic monument since 1983.
The iconic house of worship hides a lesser-known history between its walls. Over the years, the mosque became a major geopolitical asset for France â first as a way to exert control over Muslims in France during the colonial era, then during the Algerian War and later as a safehouse for Jews being persecuted under German occupation.
French historian Benjamin Stora has spent his career studying Algeria and France’s colonial past. While carrying out research on Algerian immigration and the relationship between Jewish and Muslim communities in France, he unearthed hidden truths about Paris’s Grand Mosque.
FRANCE 24: What did the Algerian community in France first think of the plan to build the Grand Mosque in Paris?
Benjamin Stora: There were very few places of worship for Muslims in Paris after World War I, and the Grand Mosque took time to be built.
Opposition to the construction project began as early as 1922 and intensified upon its inauguration in 1926, a period that coincided with the founding of the North African Star (“Ătoile Nord-Africaine” in French), the first Algerian pro-independence organisation led by Messali Hadj, who is often regarded as the father of Algerian nationalism. For the early Algerian nationalists, the mosque was seen as a tool through which the French authorities sought to counter the rise of Algerian nationalism.
Algerian immigrant workers in France were also wary of the institution, fearing it would be used to monitor, influence, or co-opt them. Over time, however, many gradually developed closer ties with the place of worship.
Did French authorities initially want to use the building to monitor Parisâs Muslim population?
At first, yes, of course. Around 100,000 Algerian workers of mostly Kabyle [a Berber ethnic group indigenous to the north of the country who mostly live in the Atlas mountains] origin arrived in France between 1919 and 1939.Many institutions were built as a way to monitor and control this population, including the Grand Mosque. There was also the Franco-Muslim Institute and the Franco-Muslim Hospital in Bobigny (a northeastern suburb of Paris). Surveillance was carried out by the North African Indigenous Affairs Department, built in 1925 and run by the Parisian police headquarters.
Because so much of the French colonial empire stretched across Muslim-majority countries, France had a lot of influence at the time. In the interwar period, the Grand Mosque was a way for France to exert that influence on the Muslim world. But it also let France project an image of tolerance toward Muslims and Islam. That ambivalence persisted throughout World War II and the Algerian War, which took place between 1954 and 1962.
Did surveillance intensify during the Algerian War?
Algerians were radically dispossessed of their land, which drove many peasants to leave the country. Itâs impossible to understand Algerian immigration without addressing this fundamental fact, especially for those of Kabyle origin.
Once in France, Algerian immigrants were forced to weather widespread segregation and lived in slums throughout the 1950s.
Surveillance in France indeed intensified during the Algerian war.It was enforced through a whole range of social and charitable organisations. The Social Action Fund, for example, was founded in 1958 and aimed at providing housing for migrant workers living in slums, but it was also used to monitor them. Â
Were Muslims wary when they visited the Grand Mosque?
Is it true that, under German occupation, the Grand Mosque was a safe haven for Jews?
French officials, judges and police officers collaborated from the beginning of the German occupation during World War II. There are photos of German army officials visiting the Grand Mosque, but they would often pose in front of institutions.
From around 1942 onwards, especially after the Velâ dâHiv roundup (a mass arrest of Jews in Paris by French Vichy police officers), the mosqueâs rector Si Kaddour Benghabrit began issuing fake Muslim IDs to Jewish children.The Jewish-Algerian singer Salim Halali, whose real first name was Simon, also took shelter in the mosque during the Occupation.
Yet there are some critics who claim that never happenedâŠ
Itâs a shame that there is controversy around this moment in the mosqueâs history. There is irrefutable evidence, such as eyewitness accounts, and I myself gathered testimonials in 2010. Itâs of course impossible to quantify just how many Jews were helped, but it definitely happened.
Some people are so bent on destroying any possibility of a friendly relationship between Jewish and Muslim communities that they want to erase this chapter in history. That erasure really saddens me.
Jews and Muslims lived together for 1,500 years. It wasnât hell, nor was it paradise, but they did coexist. Maimonides, one of the most important Jewish scholars of the Middle Ages, wrote in Arabic. Jewish prayer books were written in Arabic. Now we are in the process of destroying all of that history. Itâs important to try and preserve the places serving as bridges between religious communities. Â
The mosque has been repeatedly attacked by the right and far right in France. Why is this?
Since its construction, the Grand Mosque has been targeted by the far right. The very idea of building a mosque in Paris was hard to accept. The far right, nationalist and royalist political movement Action Française (French Action) wanted to block its inauguration. Violent campaigns and intense hostility from the far right persisted for years.
The mosque may be an undeniable part of Franceâs cultural and architectural heritage now, but it still gets caught in political crosshairs. Interior Minister Bruno Retailleaurefused to attend the Iftar meal marking the end of Ramadan in 2025 following the imprisonment of writer Boualem Sansal in Algeria. And the mosque’s current rector Chems-eddine Hafiz is constantly criticised, attacked and vilified. He faced accusations of opacity in how the mosque carried out an exclusive mandate â granted by Algeria in 2023 â to certify halal products across the EU. The attacks targeting him are in reality targeting Algeria.
There are also tensions between Algeria and Morocco around Paris’s Grand Mosque. Algeria has been accused of taking control of the site.
Morocco’s sultan may have initially headed the project, but Algeria gradually gained influence through its large population in France and the ripple effects its independence in 1962. Moroccans only began arriving in large numbers from the 1960s onwards. Itâs just a matter of history.
Welcome to our newsletter series, In Brief, where werevisit a piece from the MIT Press Reader archive and invite the author to reflect on it through a short set of questions and answers.
This month, weâre delighted to welcome Rasmus Rosenberg Larsen, Assistant Professor of Forensic Epistemology and Philosophy of Science at the University of Toronto Mississauga and author of Psychopathy Unmasked, which critically examines the scientific status and forensic application of the psychopathy diagnosis.
In a related Reader essay, Larsen argues that decades of neuroimaging research have failed to provide reliable evidence that psychopathy is associated with brain abnormalities, and that the scientific literature has often overstated what the data actually show.
You argue that the neuroscience of psychopathy is often presented as far more settled than it really is. Why do you think the image of the âpsychopathic brainâ has proved so compelling to scientists, courts, journalists, and the public?
Part of the problem is that scientists themselves have helped create the impression that a âpsychopathic brain” exists. Neuroimaging research is technically complex, and even scientifically trained readers often struggle to evaluate its methods and statistics. Lay readers therefore have little chance of recognizing when a confident conclusion is not well supported by the underlying evidence.Â
The more interesting question, I think, is why the image of the âpsychopathic brainâ became so influential within science itself. Part of the answer may lie in scientific incentives. Null findings are difficult to publish and rarely attract attention, whereas studies reporting apparent brain differences seem novel and important â and can help researchers secure lucrative grants and advance their careers.Researchers are therefore rewarded for finding and emphasizing statistically significant effects, even when the broader pattern of evidence is inconsistent or simply wrong. This dynamic is part of the wider replication and reproducibility crisis that has affected many fields, and psychopathy research strikes me as a textbook example.
But there may also be a deeper, less nefarious explanation. The traditional image of the psychopath â charming and outwardly normal, yet emotionally hollow and morally deficient â is an exceptionally powerful cultural narrative. Once researchers assume this distinct type exists, it can seem self-evident that such people must also possess distinctive brain abnormalities. That expectation may have influenced what researchers looked for, and what they believed they found.
You describe a striking gap between the actual MRI findings on psychopathy â many of them null or inconsistent â and the way those findings are summarized in review literature. How does this kind of distortion happen in science, especially when no single researcher may see themselves as acting irresponsibly?
This is a difficult question to answer because we lack conclusive evidence, so any answer must be speculative. I doubt there is deliberate or nefarious intent involved. Researchers are generally enthusiastic about their work, invested in the theories they are testing, and naturally attentive to patterns they expect to find. As I suggested earlier, the idea of the psychopath â and of a distinctive âpsychopathic brainâ â is so deeply ingrained in our culture that it may shape expectations before data are even collected.
In my view, the problem is methodological. The neuroscience of psychopathy suffers from too much analytical flexibility and too few methodological constraints. Studies often examine numerous brain regions and statistical models while relying on small or poorly controlled samples and inadequately accounting for confounding factors such as head trauma or substance use. Under these conditions, statistically significant findings are almost inevitable, even if they are spurious. That is largely what the literature shows: many null results, virtually no direct replications, and a handful of positive findings that receive disproportionate attention.
So the distortion emerges cumulatively. One exciting finding is highlighted, later reviews cite it while overlooking the many null results, and the simplified conclusion is repeated until it takes on the appearance of established fact.
What would you most like judges, lawyers, clinicians, and journalists to understand before invoking brain research in discussions of psychopathy?
The most important thing to understand is that the neuroscience of psychopathy remains wholly inconclusive. That is the central message, and it is also entirely uncontroversial. We have not established that there is such a thing as a distinctive âpsychopathic brain,â but neither have we conclusively demonstrated that no such pattern could ever exist. The evidence simply does not support any confident conclusions.
For judges, lawyers, clinicians, and journalists, the practical implication is therefore simple and straightforward: they should not invoke the âpsychopathic brainâ as though it were an established scientific finding. For now, that idea should simply be set aside.
As I show in Psychopathy Unmasked, the evidence across the major domains of psychopathy research does not support the traditional picture of the psychopath, let alone establish that such individuals even exist. The failure to identify a âpsychopathic brainâ may therefore be one symptom of a deeper mistaken assumption. Perhaps some of our most basic beliefs about psychopathy are simply wrong. There is nothing embarrassing about that. Discovering that a long-standing assumption was mistaken is not a failure of science; it is one of the most fundamental ways in which science advances.Check out these recent roundups of MIT Press Reader stories:June 2026 May 2026 April 2026 March 2026Browse previous issues of In Brief
AI drug boom
 The big picture: AI can’t replace scientific intuition. But the way it can streamline the grueling R&D before human testing begins is making computer screens as important a feature of drug design as traditional laboratories.
 Reality check: AI hasn’t yet discovered a drug that’s won FDA approval.Keep reading … Get Axios Vitals.
Dear Readers:
