Breaking AI & Tech News Analyzed

The Rise of Recursive Self-Improvement in AIThe word "recursion" is the latest buzzword in AI circles. Two separate startups have taken on the name, and many more have started referencing recursive self-improvement (RSI) in their roadmaps. Like AGI before it, RSI has become a three-letter byword for a cataclysmic AI takeoff – even if there's still a little disagreement about what it exactly means.In basic terms, RSI refers to an AI system that can continuously upgrade itself. Once AI systems can manage the upgrade cycle better than humans, the process can become a closed loop, limited only by the compute power they can access, and humans are no longer necessary or even helpful.Scary or not, that's a vision that a lot of AI labs are eager to chase.Key Players Pursuing Recursive SystemsEarlier this month, well-known AI researcher Richard Socher launched the aptly named Recursive Superintelligence with RSI as an explicit goal. "Our main focus is to build truly recursive, self-improving superintelligence at scale," Socher told TechCrunch at launch, "which means that the entire process of ideation, implementation, and validation of research ideas would be automatic."A number of other prominent researchers are already chasing that same goal, hoping for a breakthrough that will make recursive self-improvement possible.One of the most prominent is Andrej Karpathy, a legendary figure from Tesla and OpenAI, who is using agent swarms to train LLMs on simple tasks for a project he calls Auto-Research. Karpathy has been unusually open about the project, tweeting about milestones regularly and making the building blocks available through a public GitHub repo. So far, the work has mostly been confined to making minor improvements on a GPT-2 scale model — as Karpathy noted in March, "It's not novel, ground-breaking 'research' (yet)" — but it's been enough to convince lots of other researchers to follow the RSI dream. And with Karpathy now working on pre-training at Anthropic, he will have plenty of opportunity to apply the idea at a larger scale.Adaption — founded by Cohere and Google alum Sara Hooker — recently launched a similar tool called AutoScientist in an effort to automate frontier training. Like Karpathy's auto-researchers, the system trains agents to make incremental improvements — but for Adaption, the goal is to make it easier to train a full-scale frontier model. If those same researchers start to push the frontier forward, the system could quickly spiral into something very much like RSI.Disarray founder Doris Xin drew more specific RSI interest when her self-trained machine learning agent took home 28 medals in a recent Kaggle competition, beating out many human-trained agents. As she sees it, the major challenge is reliability."I would argue, given infinite compute and infinite time horizon, we are already there," Xin told me. "I want to make an argument that this is not a creative endeavor, really. It's just a lot of meat-and-potatoes engineering."The Current State of Self-Improving AIThere's also plenty of evidence that the AI industry isn't very close to recursive systems in any meaningful way — and is still grappling with talking to a wary public about its progress. So Google CEO Sundar Pichai basically admitted in a recent podcast interview."It's a continuum, and we are all definitely making progress," Pichai said. "But in the way people describe RSI, that would represent a next level of acceleration and would have a lot of implications, but we aren't quite there yet."But the continuum includes an awful lot of self-improving AI systems.In January, one of Anthropic's lead programmers for Claude Code estimated that "close to 100%" of his team's code was written by the tool — a frank admission that Claude Code was literally writing itself.Just because engineers are using an AI tool doesn't mean the tool can replace them — but Anthropic seems to be getting close to replacing engineers too. In a recent survey tied to the Mythos preview, five out of 18 Anthropic engineers believed that, with harness improvements, this version of Mythos could soon substitute for an L4 engineer — a midlevel programmer who can take on involved projects without supervision.Still, there were some of the same weaknesses you might expect."Some of Claude's major reported weaknesses compared to an L4 include: self-managing week-long ambiguous tasks, understanding org priorities, taste, verification, instruction-following, and epistemics," the report reads.In other words, its weaknesses are everything involved with self-direction, which is the cornerstone for RSI. But sure, for everything else, Claude is ready to step right in.Expert Perspectives on RSI TimelinesJust like the AGI term before it, the AI industry also can't tell us how far away it is from showcasing a meaningful recursive system. When Georgetown's Center for Security and Emerging Technology assembled a group of experts to study RSI last year, the group found a major split in assessments — some expecting an imminent "superintelligence" style explosion while others expected slower progress and an eventual plateau. But all agreed that recursion made the future especially difficult to predict.Helen Toner, director of CSET and a former board member at OpenAI, told TechCrunch that simply using AI tools to do AI research isn't enough to qualify as RSI. "They're just using AI for as much as they can," Toner told TechCrunch. "And I think that is different from the classic definition of RSI, which is really that there are no humans needed."Toner pointed to a recent post by METR's Ajeya Cotra, which distinguishes different milestones on the path to the AI research takeover. One step, which Cotra calls "adequacy," would come when the system can still perform research after all humans are removed — even if the resulting research isn't as valuable or efficient. "Parity" comes when an AI-only system is as good at research as a human-only system. "Supremacy," the final stage, comes when an AI-only system outperforms a collaborative system between humans and AI.Ultimately, Cotra concludes that AI is very close to the adequacy threshold of being able to produce some work on its own — similar to the incremental changes made by Karpathy's Auto-Research system. "I wouldn't be totally shocked if you told me this milestone had already passed, and I expect it to happen in the next couple years," Cotra wrote.She was less clear on when parity will come, but once it does, she thinks it would "massively accelerate the pace of AI progress, leading to AI research supremacy within another year."The Challenges Ahead for Recursive AIWith so much of AI built on scaling laws, there's a strong tendency to think RSI will follow the same curve. Toner thinks that many of those pursuing AI research and development via RSI "think of it as a pretty smooth ladder, where you can just keep scaling up."But even if AI researchers are able to make incremental improvements like Karpathy's auto-researchers, there will be larger challenges in handing off the whole process of research. Toner put it in terms of the history of computing, which has seen human beings handing off more and more of the process while still directing things from the top."We went from machine languages to assembly language and compiled languages; you're getting further and further from the guts of the computer," Toner said. "But the human is still, in some intuitive sense, running the show."Moving beyond that paradigm will take significant challenges, both in engineering and alignment. But even with the massive investments happening, there's no infinite compute available — and the basic trade-off between human labor and machine intelligence will be hard to overcome.The Future of Recursive Self-ImprovementAs for a total recursive AI system of apocalyptic visions? The only thing researchers essentially agree on is that, like AGI, it's not here yet.

The Lead Enterprise organizations are not rejecting AI. They are rejecting operational instability. This shift is becoming a defining reality for enterprise AI companies that scale versus those that stall after early momentum. The Event Details At TechCrunch Disrupt 2026, taking place October 13–15 at Moscone West in San Francisco, Arsalan Tavakoli-Shiraji, co-founder and SVP of field engineering at Databricks, will discuss this shift during his AI Stage session, “The Enterprise Isn’t Broken. Your Assumptions About It Are.” The Data Analysis The enterprise AI market is full of successful pilots that never became real deployments. Not because the technology failed, but because the organization could not absorb the operational consequences of adopting it. Databricks and other AI startups gaining traction inside large organizations increasingly share one thing in common: They reduce uncertainty. The Impact Analysis Enterprise buyers are asking different questions now. Concerns are no longer secondary; in many organizations, they have become core to the buying decision itself. For AI founders selling into the enterprise, understanding how technical systems interact with organizational behavior, infrastructure realities, procurement processes, governance concerns, and operational risk is crucial. The Prediction The startups that succeed in enterprise AI over the next several years may not necessarily be the ones with the most advanced models. They may be the ones that best understand how enterprises actually absorb change. The market is maturing, and enterprise AI success increasingly depends on more than strong engineering alone.

The Rise of AI-Powered Payroll Remote, a seven-year-old Amsterdam-based payroll service provider, has recently surpassed $300 million in annual recurring revenue and become cash-flow positive. However, the company's true achievement lies in its 50% increase in revenue per employee after adopting AI at every level of the organization. AI Adoption Across the Organization According to CEO Job van der Voort, the key to Remote's efficiency gains is AI adoption well beyond the CEO's office or engineering department. Employees across all functions have been launching apps in Remote Labs, an internal marketplace built on the company's own technology. The Data Behind the Growth Annual recurring revenue: over $300 million Revenue growth per employee: 50% Core payroll business growth: over 300% year over year Number of companies served: tens of thousands The Impact of AI on Remote's Business Remote's adoption of AI has not only increased revenue per employee but also improved the company's overall efficiency. The company has reduced its hiring plans and is instead focusing on upskilling its existing employees to use AI tools. The Future of AI in Payroll Remote is now opening up its AI capabilities to clients, allowing them to create custom workflows. The company has also launched Remote MCP, an interface based on the Model Context Protocol, which grants AI agents and external platforms direct access to payroll and compliance data. The Prediction As AI continues to transform the payroll industry, Remote is well-positioned to lead the charge. With its focus on AI adoption and innovation, the company is poised for continued growth and success in the future.

The Lead: Singapore's Unexpected Economic Surge Singapore's economy has grown faster than expected in the first three months of 2026, with furious demand for AI chips outweighing the fallout from the US-Israel war on Iran. The city-state's gross domestic product (GDP) expanded 6 percent year-on-year in Q1, significantly beating the official advance estimate of 4.6 percent. Technical Breakthrough: AI-Driven Manufacturing Growth On a seasonally adjusted basis, GDP grew 1 percent from the previous quarter. The Trade Ministry attributed this growth to strong performances in Singapore's wholesale trade, manufacturing, and finance and insurance sectors. In particular, robust AI-related demand led to growth in the machinery, equipment & supplies segment of the wholesale trade sector, as well as the electronics and precision engineering clusters within the manufacturing sector, the ministry stated. Financial Impact: Global Context and Regional Position Singapore accounts for approximately 10 percent of global semiconductor production and 20 percent of semiconductor chip equipment production, making it a key player in the AI revolution. The United Nations recently cut its 2026 global growth forecast to 2.5 percent (down from 2.7 percent) due to the Middle East conflict. Despite these global challenges, Singapore maintained its 2026 growth outlook at between 2 and 4 percent, acknowledging downside risks from rising energy and fertilizer prices amid the closure of the Strait of Hormuz to most shipping. Industry Transformation: The AI Boom and Singapore's Strategic Position As one of the world's most trade-reliant economies, Singapore has played a major role in the global rollout of AI technologies. The city-state's specialized manufacturing sector has benefited significantly from the ongoing AI investment boom. The AI-related investment boom is powering the manufacturing sector, and unless the Singapore economy runs out of oil, strong activity in manufacturing will continue to drive growth, said Khoon Goh, head of Asia research for ANZ. Future Outlook: Balancing Growth with Global Uncertainties Economists predict that the full impact of the Middle East crisis may become more apparent in Q2 2026, though the strong Q1 performance provides a solid foundation for the rest of the year. Local economists expect around 3.6 percent growth for 2026, acknowledging significant downside risks. The 6 percent year-on-year figure is strong, especially for a mature economy like Singapore, noted Yeow Hwee Chua, an economics professor at Nanyang Technological University. It is certainly encouraging, although I would interpret it with some caution given Singapore's high exposure to global demand and external conditions.

Lead: AI‑driven earbuds aim to transform how we sleep Boston startup SOND announced its debut product, Dreambuds, a sensor‑rich earbud that streams twelve physiological signals to a cloud‑based AI sleep coach. The launch coincides with a $7 million seed round led by MIT‑affiliated investors, positioning the company to move from prototype to mass production by mid‑2026. SOND unveils Dreambuds, a closed‑loop AI sleep earbud system Dreambuds combine high‑fidelity audio drivers with an array of sensors that monitor respiration, heart‑rate variability, cardiorespiratory coupling, sleep staging, body position, snoring, and seismocardiography (SCG). The data is processed in real time, allowing the AI coach to select or generate personalized audio programs, respond to voice commands, and adjust sleep plans without a phone. 12 physiological signals captured in‑ear Cloud AI coach with a library of 500+ audio programs Charging case includes Wi‑Fi, Bluetooth, OLED display, speaker, and physical buttons End‑to‑end operation; no smartphone required for core functions $7 million seed round backed by MIT‑linked investors The funding round was led by E14 Fund and included Crosslink Capital, Ubiquity Ventures, Alumni Ventures, Meach Cove Capital, and Boston Scientific co‑founder John Abele. The capital will finance final engineering, regulatory clearance, and a crowdfunding campaign slated for later this year. Potential shift in sleep‑tech market away from passive noise‑cancellation Traditional sleep earbuds, such as Bose’s Sleepbuds 2, focused on masking ambient noise. Dreambuds’ active, data‑driven approach could redefine consumer expectations, prompting competitors to embed richer sensor suites and AI coaching. By eliminating the need for a phone, SOND also addresses privacy concerns and user‑experience friction that have limited adoption of earlier wearables. Roadmap to mass production and market adoption by 2026‑2027 SOND plans to begin mass manufacturing in Q2 2026, following a crowdfunding round intended to raise additional runway. Early reservations are already open on the company website. If production scales as projected, Dreambuds could capture a notable share of the growing sleep‑tech market, which analysts estimate will exceed $5 billion by 2028.

The Historic Demolition EventThe towering chimneys of a decommissioned power station were dramatically brought down in a carefully planned demolition that was broadcast live to viewers around the world. The event marked the end of an era for the former energy facility, which had been a landmark in the local community for decades.Engineering Precision Behind the CollapseThe demolition required meticulous planning and execution. Engineers used strategically placed explosives to ensure the chimneys collapsed in the desired direction, minimizing damage to surrounding infrastructure. The process involved weeks of preparation, including removing hazardous materials and reinforcing structures to control the fall.Environmental Impact and RenewalThe demolition of the power station represents a significant step in the transition toward cleaner energy sources. The site is expected to be repurposed for renewable energy projects or other environmentally beneficial developments, contributing to the region's sustainability goals.Community Reaction and LegacyLocal residents watched both in person and online as the structures that once dominated the skyline came down. Many expressed mixed emotions, acknowledging the nostalgia for the industrial past while embracing the cleaner future the demolition represents.The Future of Decommissioned InfrastructureThis livestreamed demolition sets a precedent for how similar projects can be managed in the future, with increased transparency and public engagement. As more fossil fuel facilities are decommissioned worldwide, lessons from this event will inform best practices for safe and sustainable infrastructure removal.

The University of Manchester's New Initiative The University of Manchester is promising work placements to all undergraduates – regardless of their degree – to better equip them for the challenges of the current job market. This move appears to be a first for a large Russell Group institution. Details of the Placement Program Manchester’s vice-chancellor, Duncan Ivison, emphasized that no student should graduate having done three years of just academic study. Instead, “every single student [should] have a chance to put their learning into context – an internship, a placement, a joint project or an exchange”. The program aims to provide “meaningful real-world experience” to all students, from classics to chemical engineering. The initiative includes placements, short internships, live employer projects, or work with public or community organizations. The Data Analysis The plan comes as graduates increasingly struggle to find work after leaving university, some with debts of more than £50,000. Those who do get work are often in low-paid roles in hospitality or retail, rather than traditional graduate jobs. 32,000 undergraduates are currently enrolled at Manchester University. In 2024-25, almost a quarter of undergraduate courses gave students the option of a placement of at least a year. The Impact Analysis Nick Hillman, the director of the Higher Education Policy Institute, welcomed the initiative but raised feasibility concerns due to the large number of students and employers involved. He noted that some universities, such as Aston and Loughborough, have always embedded employment into their courses. The Prediction Vivienne Stern, chief executive of Universities UK, welcomed the initiative, stating that the jobs market is changing rapidly and universities have an important role to play in preparing students for the world of work. Libby Hackett, the chief executive of the Russell Group, also supported the move, highlighting its significance in equipping graduates to navigate the changing workplace.

The Graduation Backlash Against AI OptimismWhen Jacob Pagel graduated from Middle Tennessee State University this spring, predictions about artificial intelligence already had him questioning the value of his degree. Then a music executive started preaching about AI's transformative power during a commencement speech."This industry will change on you in a heartbeat. It has already changed more in the last 10 years than in the 50 years prior … AI is rewriting production as we sit here," said Scott Borchetta, CEO of the record label Big Machine. After a few stray boos from graduates, he doubled down: "Deal with it."The students' jeering grew louder, but Borchetta barreled through: "You can hear me now or you can pay me later … then do something about it. It's a tool. Make it work for you." He continued: "The things you learned in your first year here may already be obsolete."Multiple Universities, Same Student FrustrationBorchetta's speech is one of several at commencement ceremonies this spring that have revealed a disconnect between the executives championing AI and students, eliciting derision in real time even for Google's former CEO. Recent graduates at the University of Central Florida and the University of Arizona booed speakers who compared the advent of AI to the Industrial Revolution and the development of the laptop and smartphone.At the University of Arizona, 20-year-old Arian Chavez, was angry about his school's decision to let ex-Google CEO Eric Schmidt speak, even before he got on stage. Chavez, a junior studying chemical engineering, is part of a group called Students for Socialism, and helped them organize an online petition to remove Schmidt as a commencement speaker."I know what many of you are feeling about that. I can hear you," Schmidt said, amid a chorus of boos. "There is a fear in your generation that the future has already been written, that the machines are coming, that the jobs are evaporating, that the climate is breaking, that politics is fractured, and that you are inheriting a mess that you did not create, and I understand that fear."Public Sentiment: AI's Poor ReceptionThe students at these ceremonies "are a mouthpiece for the population at large", according to Cornell University professor Sarah Kreps, who has studied societies' reactions to new technology. "These tech executives are not reading the room … These kids have spent hundreds of thousands of dollars on a degree that they don't know will serve them well."While they may feel AI's disruptive effects acutely as entry-level job seekers, AI has proved unpopular among the general US public. A national survey conducted for NBC News earlier this year polled 1,000 registered voters and found only 26% view AI positively and 46% view it negatively. AI scored worse than US Immigration and Customs Enforcement (ICE), Donald Trump and Kamala Harris on the same poll, but better than the Democratic party and Iran.Anger against AI is palpable across the country – from communities protesting against datacenters powering the AI boom, to workers disputing their CEOs' claims that AI can, effectively, replace them.The Economic Reality Behind the Student AnxietyPagel and his peers are entering a job market where AI's efficiency is already being used to justify mass layoffs. While it's unclear which jobs may be entirely replaced by AI – and whether AI could eventually create more career pathways than it destroys – recent graduates are feeling betrayed."We've been pushed our entire lives to get our diplomas. Then you pulled the rug out from underneath us, and said: 'Oh, you know those four years you spent learning how to do very specific things, you don't need to do it any more,'" Pagel says. "We can get a computer to do it for two-thirds the price."CEOs' graduation speeches about AI have become a preventable PR disaster, according to Parry Headrick, founder of Crackle PR, a tech public relations agency that has worked with startups. Executives should have acknowledged and reassured students' anxieties, while also advising them to adapt."What in the heck is anybody who is young and in school supposed to do when you have these tech executives beating their chests about the next Industrial Revolution when they can't afford to buy groceries or pay for rent?" Headrick asks. Nearly half of college students said their financial stress made it hard to concentrate on their coursework, according to a 2026 report from Trellis Strategies, a research group focused on postsecondary education.AI's Practical Impact on Education CeremoniesAt Glendale Community College in Arizona, it wasn't a graduation speaker that drew students' ire, but the AI-powered machine reading out their names. Turns out, it missed some.College president Tiffany Hernandez apologized and told graduates towards the end of the ceremony: "Here's what's happening. We're using a new AI system as our reader," she said, as boos roared through the arena. Hernandez paused for a few seconds and let out a few nervous laughs. "That's a lesson learned from us."Aidan Benjamin, who is graduating from Glendale Community College this summer with an associate's degree in accounting, was at the ceremony to support his cousin. He thought she would be walking the stage. She never did, because the AI announcement system never called her name."I was booing because I was like, this sucks. This is such a big moment for students." Benjamin said they both laughed about the malfunction afterwards. "But it just didn't feel good at the end of the day, like, it shouldn't have happened that way," he says.The Future of AI in Education and CareersPagel is considering a career in helping children undergoing medical treatment, or entering politics – perhaps running for office, or working as a liaison for federal agencies. "That sphere depends on human face-to-face interaction. No computer can take that," he says, calling AI-generated campaign ads "the cheap route"."It's up to us as engineering students to use our knowledge for the service of the planet and not billionaires," says Arian Chavez, who wants to work in the environmental regulation of chemical plants.As AI continues to reshape industries and education, the graduation protests may represent an early indicator of a generational shift in how technology is perceived – not as an unqualified good, but as a force that requires careful management to avoid displacing workers and devaluing human expertise.

The Plight of Immigrant Students in France France is grappling with a concerning trend of deporting immigrant students who have graduated from technical high schools. In the northern suburb of Paris, Saint-Denis, dozens of students of immigrant backgrounds are facing deportation orders, despite exemptions designed to protect students training in understaffed fields. Deportation Orders on the Rise In April, Mariem*, a 19-year-old Tunisian student studying medicine, received a deportation order. Mohammed*, 19, a Moroccan student enrolled in an electrical engineering program, also received a similar letter. The deportation orders, known as "Obligation de Quitter le Territoire Francais" (OQTF), have left many students in a state of shock and uncertainty. The Data Analysis An estimated 50 students in one Saint-Denis trade school were either undocumented or had been deported due to paperwork issues. About a dozen other technical schools in the region are experiencing similar challenges. The percentage of visas issued overall has decreased by more than 40 percent, and work visas by more than 50 percent. The Impact Analysis The deportation orders are often the result of administrative errors, convoluted bureaucracy, or language barriers. Many students have lost out on work or study opportunities due to the prefecture's timeline to approve work permits falling outside the academic calendar. The backlog for residence and work permits is particularly bad in Saint-Denis, with an average processing time of 145 days, compared to 117 across France. The Prediction The situation may improve with the election of a left-wing mayor in Saint-Denis, Bally Bagayoko, who has shown support for undocumented technical school students. However, the road ahead remains uncertain for many immigrant students in France.