ON

📰 MIT Technology Review · 33

United StatesBusiness
2 days ago

A startup claims it broke through a bottleneck that’s holding back LLMs

A Miami-based AI startup named Subquadratic has claimed to have overcome a major mathematical bottleneck limiting large language models (LLMs). The company introduced a new model called SubQ, which it says is faster, cheaper, and more energy-efficient than existing models. SubQ is reported to handle significantly more text at once and perform well on key tasks compared to leading models from companies like Google DeepMind, OpenAI, and Anthropic. However, initial skepticism existed due to limited evidence, though the company has since shared results from an independent evaluation.

0 left1 center0 right
3 sources2 officialviaTechCrunch
United StatesBusiness
2 days ago

The Download: AI bottleneck debates, and BCI trials take off

The article discusses recent advancements in artificial intelligence, focusing on a new transformer-based model called Subquadratic that claims to reduce computational requirements, making large language models faster and more energy-efficient. While some experts remain skeptical, the company is beginning to share details about their approach. Additionally, the article highlights the growing interest in brain-computer interfaces (BCIs), particularly through the case of Casey Harrell, an ALS patient using a brain implant to maintain independence. The piece also mentions China's approval of a BЦ

0 left0 center0 right
1 source3 officialviaMIT Technology Review
United StatesScience
2 days ago

The inevitable weakness of metrics

The article discusses the dual nature of metrics, highlighting their ability to provide insight while also potentially obscuring or distorting reality. The author reflects on their personal experience with self-tracking, noting initial motivations centered around gaining self-knowledge rather than optimization. The piece explores how metrics can create a false sense of clarity and control.

0 left0 center0 right
1 sourceviaMIT Technology Review
United StatesMedicine
2 days ago

Fake MAGA? Anti-AI Group’s Conservative Branding Crumbles

An organization called Humans First, which presents itself as a conservative group opposing the expansion of AI data centers, has been revealed to have ties to leftist activists. The group's leader, Amy Kremer, a former Tea Party figure associated with the January 6 events, frames the movement as a grassroots conservative response to perceived elite control by Silicon Valley. However, investigations show that key members of Humans First have backgrounds in leftist activism, including work with the Sunrise Movement, a group known for opposing Trump and advocating for aggressive climate action.

2 left2 center1 right
8 sources8 officialviaThe Daily Wire
United StatesCulture
2 days ago

Trump's fight with Anthropic is now a fight over cybersecurity

The Trump administration's restrictions on Anthropic's AI models have raised concerns among cybersecurity experts about potential negative impacts on U.S. cyber defenses. Prominent security leaders, including former Facebook security chief Alex Stamos, have criticized the administration's actions, warning that they could discourage American AI companies from developing tools that help identify and fix vulnerabilities. An open letter signed by nearly 150 security professionals called for the reversal of these restrictions. The dispute arose following Anthropic's release of its first Mythos-5 AI

2 left21 center4 right
15 sources37 officialviaBloomberg News
United StatesEconomy
2 days ago

SpaceX shoots 19% higher in first day of trading

SpaceX shares rose 11% in initial trades, giving the company a market capitalization of $1.97 trillion and placing it among the top 10 largest U.S. companies. The stock was priced at $135 before trading began, and if included in the S&P 500, it would represent more than 5% of the index. Elon Musk expressed skepticism about SpaceX’s chances of success prior to the IPO. Analysts note that SpaceX’s valuation is based on sales rather than profits, raising concerns about whether it aligns with fundamental business metrics. However, similar high valuations are common among top-performing S&P 500 equ

0 left8 center0 right
23 sources14 officialviaQuartz
United StatesBusiness
3 days ago

The search for dark matter has been blown wide open

Physicists have been searching for dark matter for decades, using underground detectors filled with liquid xenon to detect potential collisions between dark matter particles and xenon atoms. Recent findings suggest that the signals detected are likely caused by neutrinos rather than dark matter. This has prompted researchers to explore alternative methods for detecting dark matter, including quantum sensors and experiments in Jupiter's atmosphere.

0 left0 center0 right
1 source1 officialviaMIT Technology Review
United StatesScience
3 days ago

Hacking the atmosphere: Geoengineering gets a reality check

The article discusses the development of specialized aircraft designed to operate in the stratosphere for the purpose of solar geoengineering—a controversial approach to mitigating global warming by reflecting sunlight back into space. The technology involves releasing materials at high altitudes to scatter sunlight, inspired by the natural cooling effects observed after major volcanic eruptions. Jim Franke, a researcher at the University of Chicago, highlights the engineering challenges involved in achieving such high-altitude operations.

0 left2 center0 right
1 source4 officialviaMIT Technology Review
United StatesSports
4 days ago

Entrepreneurs in Nairobi make the case for going solar

Entrepreneurs in Nairobi are promoting off-grid solar solutions to address the lack of centralized electricity in parts of Kenya. With 25% of communities without access to the national grid, Kenya aims to achieve universal electricity access by 2030 while avoiding increased emissions. Solar technology has become more economically viable, with panel costs dropping significantly over recent years. One example is Milcah Wanjiru, who operates a small shop where she provides a milling service powered by solar energy or electricity.

0 left0 center0 right
1 source1 officialviaMIT Technology Review
United StatesEconomy
4 days ago

OpenAI files confidential SEC paperwork for IPO, opening the door to a Wall Street debut

OpenAI has filed preliminary, confidential paperwork with the U.S. Securities and Exchange Commission (SEC), signaling its potential move toward an initial public offering (IPO). This development positions OpenAI as part of a growing group of major artificial intelligence companies preparing for Wall Street debuts, including rivals like Anthropic and SpaceX. OpenAI CEO Sam Altman previously indicated that an IPO could be the most viable path for the company due to its size and the substantial capital required for technological advancement. The company emphasized that no final decisions haveyet

1 left11 center1 right
13 sources21 officialviaAxios
United StatesMedicine
4 days ago

Exclusive eBook: How AI is becoming the next military advisor

MIT Technology Review has released an exclusive subscriber-only eBook exploring how artificial intelligence is being integrated into military decision-making processes. The eBook compiles six previously published articles, updated with recent developments, covering the use of AI models by militaries.

0 left0 center0 right
1 sourceviaMIT Technology Review
United StatesSports
5 days ago

Want to get a data center online quickly? Give it some flex.

The article discusses a simulated scenario where an AI system, called Conductor, was tested to manage electricity demand by adjusting the operations of a data center. The simulation recreated conditions similar to a high-demand event in the UK, such as a major sports event causing a surge in household energy usage. The goal was to explore whether data centers could become more flexible with their power consumption to support grid stability.

0 left0 center0 right
1 source2 officialviaMIT Technology Review
United StatesCulture
5 days ago

Why do South Koreans love AI so much?

The article describes the author's experience upon arriving in Seoul, highlighting various instances of advanced technology and AI integration in daily life, such as facial recognition at immigration checkpoints, 5G-enabled smartphones, AI-powered bus stops, and the general enthusiasm of South Koreans toward AI. It contrasts this optimism with the growing public concern over AI in the United States.

0 left0 center0 right
1 sourceviaMIT Technology Review
United StatesEconomyOverlooked from the right
5 days ago

Visa plugs its payment network into ChatGPT, letting AI agents shop and pay for users

Payments giant Visa has integrated its payment network into ChatGPT, allowing the AI chatbot to shop and complete transactions on behalf of users. This partnership enables AI agents to recommend products and make purchases at any merchant that accepts Visa. Unlike OpenAI's earlier Instant Checkout feature, which was discontinued due to adoption issues, Visa's integration aims to streamline the process for both users and merchants by linking Visa cards directly to ChatGPT.

1 left0 center0 right
6 sources6 officialviaQuartz
United StatesEnvironment
6 days ago

The Download: cutting AC emissions, and nature’s drug designer

The article discusses advancements in solid-state cooling technologies aimed at reducing the environmental impact of air conditioning, which currently contributes significantly to global energy consumption and greenhouse gas emissions. It also highlights chemist Tim Cernak's shift from pharmaceutical work to developing natural alternatives for animal healthcare.

0 left0 center0 right
1 sourceviaMIT Technology Review
United StatesScience
6 days ago

These new solid-state ACs promise a cool future. Scientists aren’t so sure.

Scientists and startups are exploring solid-state cooling technologies as an alternative to traditional air conditioning systems. Solid-state cooling uses conductive materials such as gadolinium and bismuth telluride to transfer heat without compressors or fans. While promising, these systems face challenges in matching the efficiency of conventional methods. Research and pilot projects are ongoing, including efforts by companies like Mimic Systems and Magnotherm.

0 left0 center0 right
1 source1 officialviaMIT Technology Review
United StatesBusiness
9 days ago

Why “reprogramming” is the buzziest approach to reversing aging right now

Life Biosciences, a biotechnology company, has begun testing an experimental treatment for glaucoma by injecting a therapy directly into the eye to regenerate healthy nerve tissue. The treatment aims to reverse damage caused by the disease and potentially extend to other age-related conditions. The approach involves 'reprogramming' cells to a younger state, which is currently a prominent strategy among biotech firms exploring ways to slow or reverse aging. Researchers have identified several key biological processes associated with aging, though some areas of focus have evolved over time.

0 left0 center0 right
1 source2 officialviaMIT Technology Review
United StatesCulture
9 days ago

You do your own time

The article opens with a vivid scene set in a library where characters including 'Little Jo,' 'Eustace,' and the narrator prepare for an unexpected visitor. The narrative describes the arrival of a stranger seeking sanctuary, with detailed descriptions of actions taken by the characters to assist the individual.

0 left0 center0 right
1 sourceviaMIT Technology Review
United StatesBusiness
9 days ago

Inside interoception: The hidden sense of how you feel inside

The article discusses interoception, the internal sense that allows the brain to perceive bodily states such as heart rate, temperature, and emotional responses without conscious awareness. It explains how the brain processes vast amounts of sensory input unconsciously, filtering out most information to focus on what is necessary for survival and decision-making. Neuroscientist Moriah Thomason comments on the importance of this unconscious processing.

0 left0 center0 right
1 sourceviaMIT Technology Review
United StatesSports
10 days ago

The Download: soccer’s data renaissance and China’s big nuclear plans

The article discusses two main topics: the use of data analytics and AI in transforming soccer strategy, and China's rapid expansion of large nuclear reactor projects. It highlights research by Professor Jesse Davis and his team at KU Leuven, who are using advanced analytics to uncover new tactical approaches in soccer. Additionally, it covers China's growth in nuclear energy, noting the country's significant increase in nuclear power capacity.

0 left0 center0 right
1 source2 officialviaMIT Technology Review
United StatesScience
10 days ago

Gen Z’s Ability to Detect AI Is Far Lower Than You’d Expect

A new poll indicates that young people are overconfident in their ability to detect AI-generated content. The article discusses the release of Google DeepMind's Veo 3 AI video generator and highlights concerns about the proliferation of generative AI tools, including their potential misuse by bad-faith actors. It references the spring 2026 Yale Youth Poll, which tested participants' ability to distinguish between real and AI-generated images.

0 left1 center0 right
2 sources5 officialviaMIT Technology Review
United StatesBusiness
10 days ago

Job titles of the future: Nature’s drug designer

Chemist Tim Cernak transitioned from pharmaceutical research to using his expertise to develop treatments for animals affected by diseases. He uses AI tools such as AlphaFold to design drugs tailored specifically for non-human patients, aiming to minimize ecological impact.

0 left0 center0 right
1 source1 officialviaMIT Technology Review
United StatesEconomy
10 days ago

Why China is betting on big nuclear reactors

The article discusses China's focus on large-scale nuclear reactors as part of its strategy to meet growing electricity demands while avoiding greenhouse gas emissions. It contrasts this approach with the United States and France, which have struggled to add new reactors recently. The U.S. has only completed two reactors at Plant Vogtle in Georgia, while France's most recent reactor was connected to the grid in December 2024—the first in over 20 years. The article highlights the challenges of building large nuclear projects, including high upfront costs, design complexity, and delays due to监管.

0 left1 center0 right
1 source1 officialviaMIT Technology Review
United StatesSports
10 days ago

Inside soccer’s data renaissance

The article discusses the growing role of data analytics in soccer, focusing on the work of Professor Jesse Davis and his team at KU Leuven's Sports Analytics Lab. The lab uses machine learning to analyze soccer strategies, helping professional clubs make better decisions regarding player evaluation, strategy efficiency, and identifying hidden tactical patterns.

0 left0 center0 right
1 source2 officialviaMIT Technology Review
United StatesSports
10 days ago

The “steroid olympics” were a circus—and a window into our culture

The article describes the Enhanced Games, a sporting event held in Las Vegas where athletes are encouraged to use performance-enhancing drugs. It highlights the variety of substances used, such as testosterone, human growth hormone, and others, and discusses both the organizers' goals—challenging traditional sports norms and promoting advancements in human health—and critics' concerns about the dangers and ethical issues involved.

0 left1 center0 right
2 sources1 officialviaTechCrunch
United StatesBusiness
12 days ago

The Download: whole-body rejuvenation drugs and five things to know about AI

MIT Technology Review reports on two main topics: David Sinclair's plan to test whole-body rejuvenation drugs in the XPrize competition and a summary of key AI trends presented at SXSW London. Sinclair aims to demonstrate 'age restoration' through a drug regimen, with the goal of improving immune, cognitive, and muscle functions by up to 10 years within a year of treatment. The article also highlights recent developments in artificial intelligence.

0 left0 center0 right
1 source2 officialviaMIT Technology Review
United StatesTechnology
12 days ago

Learning to lead in a hybrid human-AI enterprise

The article discusses the rapid growth of AI agent adoption in enterprises, predicting a 300% increase over the next two years. It highlights how these AI agents differ from traditional automation by operating autonomously and collaborating with humans in various departments such as customer service, HR, and sales. The piece notes significant productivity improvements ranging from 30-50% in early implementations. It also emphasizes the anticipated transformation of workplace dynamics and the need for leadership to adapt to managing digital labor shaped by agentic AI.

0 left0 center0 right
1 source1 officialviaMIT Technology Review
United StatesBusiness
12 days ago

David Sinclair plans to test whole-body rejuvenation drugs in the XPrize competition

David Sinclair, a longevity scientist at Harvard Medical School, plans to conduct human trials of an oral 'reprogramming' drug as part of the XPrize Foundation's $101 million competition aimed at restoring individuals to an earlier apparent age through improvements in immune, cognitive, and muscle function. The goal is to demonstrate evidence of age restoration in humans.

0 left0 center0 right
1 source1 officialviaMIT Technology Review
United StatesTechnology
12 days ago

Five things you need to know about AI

In a talk at SXSW London, MIT Technology Review senior editor Will Douglas Heaven discussed five key themes in AI as of mid-2026. He noted that generative AI tools have become commonplace, automating many office tasks, including creating presentations. The discussion highlights ongoing concerns about the impact of AI on employment, though the implications remain unclear. Heaven also mentioned that much has changed since his similar talk at SXSW London the previous year.

0 left0 center0 right
1 sourceviaMIT Technology Review
United StatesCulture
12 days ago

The Download: AI hacking beyond Mythos, and chatbots’ impact on our brains

The article discusses two main topics: an AI-related security incident involving Meta's customer support agent and research suggesting AI chatbots might negatively affect human cognition. In the first section, it describes how attackers exploited Meta's AI system to steal Instagram accounts by asking the AI to link accounts to controlled email addresses. This highlights vulnerabilities in AI systems beyond just advanced models like Anthropic's Mythos. In the second section, psychologist Gloria Mark raises concerns that AI chatbots could contribute to declining attention spans and increased-st

0 left3 center0 right
2 sources5 officialviaAssociated Press
United StatesSports
13 days ago

Why this year’s World Cup ball may not fly as far

A team of researchers has analyzed the aerodynamics of Adidas's new World Cup ball, the Trionda, which features unique panel designs and emblems representing the host countries. Wind tunnel tests suggest that while the ball may slightly reduce the distance of long kicks, it could improve accuracy and predictability of flight. The findings highlight potential impacts on players' strategies, particularly for goalkeepers and long-range shooters.

0 left0 center0 right
1 source1 officialviaMIT Technology Review
United StatesScience
17 days ago

How virtual power plants could provide energy for data centers

Voltus is implementing a virtual power plant (VPP) that aggregates energy resources such as electric vehicles and smart thermostats. Participants will be compensated for contributing to the system, which will manage energy usage during peak times. Google is funding the initiative, and the additional energy capacity will support its data centers. This represents a significant step toward using VPPs to address energy demands for large-scale operations like data centers. However, challenges remain regarding the scalability and limitations of such programs. Research suggests that data centers can暂

0 left0 center0 right
2 sources1 officialviaSemafor
United StatesCrime
17 days ago

How courts are coping with a flood of AI-generated lawsuits

The article discusses how U.S. courts are dealing with an increase in lawsuits filed by individuals representing themselves, which some judges attribute to the use of AI tools. Judge Maritza Braswell notes that AI is helping unrepresented litigants draft better legal documents but does not necessarily improve their chances of success. The piece raises questions about the ethical and legal implications of using AI in place of human lawyers.

0 left0 center0 right
1 source1 officialviaMIT Technology Review