A homegrown version of the extreme ultra-violet lithography system needed to produce the most advanced chips is on trial at a Huawei facility, say reports.
Richard Platt's insight:
Yes, Chinese foundries can knock out less sophisticated chips with fewer transistors. Through techniques such as double and quadruple patterning, they might even be able to produce ICs with transistor measurements of just 7 nm's. SMIC, China's best-known foundry, appeared to have made one for the Mate 60 Pro, a Huawei smartphone released in 2023. In layman's terms, EUV works by firing a concentrated beam of light off the smoothest mirrors in the world and onto a silicon wafer, where it etches complex circuit designs like an Antman scribe. It uses a wavelength measuring just 13.5 nm, compared with the 193 nm of deep ultraviolet lithography (DUV), its predecessor. The difference is like that between a fat marker and a thin ballpoint to a fine artist. According to various reports China's Huawei, develped an EUV system called laser-induced discharge plasma (LDP) technology that's been going through tests at a Huawei facility in Dongguan. One report says it has been able to generate the 13.5-nanometer wavelength by "vaporizing tin between electrodes and converting it to plasma via high-voltage discharge, where electron-ion collisions produce the required wavelength." "It's a pretty cool technique because it's actually simpler than what ASML does," said Earl Lum, a IC expert at EJL Wireless Research. "It could be cheaper to make the machine because of the strategy that ASML had to use." Trials do not mean a Chinese flavor of EUV is close to commercial deployment, of course, and a few press reports that leave many questions unanswered must be treated with a generous dose of skepticism. ASML's share price fell 7% on March 10, days after the reports about China's apparent EUV breakthrough. But this may have been linked to more general concerns about tariffs and their economic impact. Other chip stocks also suffered.
On this first episode of "Bloomberg Tech: Asia," we explore the impact of China's AI advancements on competition with the US. We hear from key players in Asia's AI revolution including Manycore, one of China's so-called "Six Dragons" and Tokyo Electron, one of the world's top chip-tool makers. China Growth Capital's Wayne Shiong and Power Dynamics' Jen Zhu Scott also weigh in on the outlook for the AI race and investments in the sector.
AI has already infiltrated the workforce, so higher ed institutions have a responsibility to teach their students to use it responsibly and effectively.
AI has already infiltrated the workforce, so higher ed institutions have a responsibility to teach their students to use it responsibly and effectively. Without clear guidance, training, and inclusion, many Gen Zers risk being left behind in an AI-driven economy. Schools and employers must step up by creating inclusive policies, integrating AI education, and expanding access to tools and training, especially in underserved sectors and communities.
Without clear guidance, training and inclusion, many Gen Zers risk being left behind in an AI-driven economy. Schools and employers must step up by creating inclusive policies, integrating AI education and expanding access to tools and training, especially in underserved sectors and communities.
Artificial intelligence is no longer a distant possibility. It is a defining reality of our present moment. From predictive analytics in admissions to generative AI tools shaping classroom practice and research workflows, AI is rapidly transforming higher education. Yet this transformation is not simply technological. It is cultural, ethical, and institutional. The question before us is not whether we will use AI but whether we will guide its use with purpose, clarity, and care.
Artificial intelligence is no longer a distant possibility. It is a defining reality of our present moment. From predictive analytics in admissions to generative AI tools shaping classroom practice and research workflows, AI is rapidly transforming higher education. Yet this transformation is not simply technological. It is cultural, ethical, and institutional. The question before us is not whether we will use AI but whether we will guide its use with purpose, clarity, and care.
Recent generations of frontier language models have introduced Large Reasoning Models (LRMs) that generate detailed thinking processes…
Richard Platt's insight:
Recent generations of frontier language models have introduced Large Reasoning Models (LRMs) that generate detailed thinking processes before providing answers. While these models demonstrate improved performance on reasoning benchmarks, their fundamental capabilities, scaling properties, and limitations remain insufficiently understood. Current evaluations primarily focus on established mathematical and coding benchmarks, with an emphasis on final answer accuracy. However, this evaluation paradigm often suffers from data contamination and does not provide insights into the reasoning traces’ structure and quality. In this work, we systematically investigate these gaps using controllable puzzle environments that allow for precise manipulation of compositional complexity while maintaining consistent logical structures. This setup enables the analysis of not only final answers but also the internal reasoning traces, offering insights into how LRMs “think”. Through extensive experimentation across diverse puzzles, we show that frontier LRMs face a complete accuracy collapse beyond certain complexities. Moreover, they exhibit a counter-intuitive scaling limit: their reasoning effort increases with problem complexity up to a point, then declines despite having an adequate token budget. By comparing LRMs with their standard LLM counterparts under equivalent inference compute, we identify three performance regimes: (1) low-complexity tasks where standard models surprisingly outperform LRMs, (2) medium-complexity tasks where additional thinking in LRMs demonstrates advantage, and (3) high-complexity tasks where both models experience complete collapse. We found that LRMs have limitations in exact computation: they fail to use explicit algorithms and reason inconsistently across puzzles. We also investigate the reasoning traces in more depth, studying the patterns of explored solutions and analyzing the models’ computational behavior, shedding light on their strengths, limitations, and ultimately raising crucial questions about their true reasoning capabilities.
Startup EnCharge aims to solve the analog compute problem with metal-layer capacitors that help boost energy efficiency and performance.
Richard Platt's insight:
Analog computing is not a new idea, but the emergence of math-heavy AI workloads in recent years has prompted several startups to build new architectures based on some of the same concepts. In general, the basic operations of multiplication and addition are achieved within a memory array. A memory cell stores a weight, acting as a variable resistor with resistance in some way proportional to the weight value. Data is encoded onto a voltage, which, when supplied to the memory cell, effectively multiplies the data value by the weight value. Output wires are joined together such that currents combine as a simple form of addition. This is a very low-energy way to multiply and add, the two math operations required for matrix multiplication, which form the bulk of AI workloads. Having computation take place in the memory—where the weights are already stored—also means less data movement is needed, which is more energy efficient. Other companies’ analog computing schemes have had various levels of success over the years. Mythic uses an array of Flash memory cells as a matrix multiply accelerator, for example, but this requires complex calibration algorithms for process and temperature variations that can reduce precision. Other types of memory can be used; Tetramem uses RRAM in its memory array. D-Matrix uses modified SRAM for analog multiply, combined with digital addition in its scheme to get around problems with precision and accuracy in all-analog designs.
Using a newly discovered byproduct of dying cancer cells, University of Wisconsin–Madison researchers are developing personalized vaccines that could help keep aggressive tumors from recurring.
Richard Platt's insight:
Using a newly discovered byproduct of dying cancer cells, University of Wisconsin–Madison researchers are developing personalized vaccines that could help keep aggressive tumors from recurring.
The personalized vaccine approach is an extension of the team's recent discovery of pyroptotic vesicles, which are tiny sacs filled with the remnants of cancer cells when they undergo programmed cell death. Crucially, the remnants in these microscopic sacs include antigens specific to the tumor, along with other molecular bits that can help direct immune cells to find and suppress cancer cells that might remain after a tumor is surgically removed. In their study, recently published in the journal Nature Nanotechnology, Hu and his colleagues engineered these sacs to carry an immune-stimulating drug. They then embedded these engineered vesicles into a hydrogel that can be implanted into the space left behind after surgical removal of a tumor. Using a melanoma mouse model and two different types of mouse models for triple negative breast cancers, including one with a human-derived tumor, the researchers compared their new approach with other cancer vaccine methods being studied. The mice that received the hydrogel laden with their engineered sacs survived significantly longer than others "Compared to the other approaches, ours shows a much stronger immune response," says Hu. "We were one of the first groups to identify these pyrotopic vesicles and the first to show their effectiveness in helping prevent cancer recurrence, and we are very excited about their potential."
Led by Quanyin Hu, a professor in the UW–Madison School of Pharmacy, the research team has already found success slowing the recurrence of tumors in mouse models of triple negative breast cancer and melanoma. Currently, the long-term prognosis for human patients with these cancers is relatively poor. That's in part because the diseases have a tendency to recur after the initial treatments to remove the tumors.
News on dementia trends, health care technology, vaccines, and other health topics....
Richard Platt's insight:
Rising Numbers: An estimated 7.2 million Americans age 65 and older are living with Alzheimer’s dementia, or roughly 1 in 9 older adults. The health and long-term care costs for Alzheimer’s and other dementias are significant, projected to reach $384 billion in 2025. This does not include an additional estimated $413.5 billion in unpaid caregiving, often provided by family members and friends.
New Risk Factors: Recent research highlights several modifiable risk factors for Alzheimer’s, including sedentary behavior, type 2 diabetes, and reduced sleep quality. Herpes simplex virus 1 infections (cold sores) or high cortisol levels (specifically in post-menopausal women) may also contribute to the increased risk of dementia due to changes in the brain.
Recent Advancements: While there is no cure for dementia, the latest advancements may make earlier diagnosis more accessible and less invasive. The Food and Drug Administration (FDA) recently approved the first blood test to detect Alzheimer’s-related brain changes. Researchers have also found new potential pathways through which Alzheimer’s and other dementias develop, which could contribute to future treatment options.
We see increasing levels of disengagement from the curriculum. Fewer students carry on to higher education. The intellectual elites become smaller and more powerful, but we also see a disruption. Academia is peeled away. Innovation occurs outside of the walls of schools.
We see increasing levels of disengagement from the curriculum. Fewer students carry on to higher education. The intellectual elites become smaller and more powerful, but we also see a disruption. Academia is peeled away. Innovation occurs outside of the walls of schools. -- This article from the AI English Teacher looks at how we can educate students in the future to ensure that we aren’t just evaluating their use of AI and also touches on why this probably won’t happen. Can you guess why? - Well worth reading https://theaienglishteacher.wordpress.com/2025/06/14/two-futures-a-choice-for-education-in-the-age-of-ai/
Clinical decision-making in oncology is challenging and requires the analysis of various data types – from medical imaging and genetic information to patient records and treatment guidelines.
Clinical decision-making in oncology is challenging and requires the analysis of various data types – from medical imaging and genetic information to patient records and treatment guidelines.
Purpose The use of Artificial Intelligence (AI) in education has the potential to further customise and personalise students’ learning, and encourage self-directed learning. It can also augment teachers’ professional practice by automating routine tasks and allowing teachers to spend more time...
Purpose The use of Artificial Intelligence (AI) in education has the potential to further customise and personalise students’ learning, and encourage self-directed learning. It can also augment teachers’ professional practice by automating routine tasks and allowing teachers to spend more time...
AI, ChatGPT, and LLMs "have absolutely blown up what I try to accomplish with my teaching." Nik Peachey's insight Some interesting comments from teachers in this article about how AI has now impacted their teaching
In a landmark comparative study published in the Journal of Health Organization and Management, researchers from the University of Maine have embarked on a rigorous investigation to evaluate the diagnostic capabilities of artificial intelligence (AI) models against those of seasoned human clinicians...
In a landmark comparative study published in the Journal of Health Organization and Management, researchers from the University of Maine have embarked on a rigorous investigation to evaluate the diagnostic capabilities of artificial intelligence (AI) models against those of seasoned human clinicians.
The Chinese tech sector has been on a roll since the arrival in January of DeepSeek, the AI startup that stunned the world with a language model that claimed to match or outperform Western rivals, at a fraction of the cost. Bloomberg's Annabelle Droulers reports on how the rapid strides in AI are poised to escalate the tech "cold war" between the US and China.
Last week in my Sunday Suggestions I shared a prompt that turned your AI chatbot into a critical friend as an example of how AI CAN develop critical thinking, we just have to think a little more critically about how we apply it!
However, I suspect that what many teachers mean, when they talk about the impact of AI on critical thinking, is that students will get information from AI sources and won't think critically about whether or not it's true.
The widespread investment in AI furthers economists’ optimism about a “roaring 20’s” of worker productivity on the horizon. However, this will not take place in health care without accompanying systemic and organizational actions that rethink what we financially incentivize, how we integrate new technologies, how we shift tasks, and how we prepare the workforce.
Two years ago, every health conference I attended had multiple panels on clinician burnout. The problem is well known, and the actual system-level contributors have been called out.
In the last year, the solution to burnout is splashed on every conference app log-in screen: generative artificial intelligence (AI).
The AI scribe revolution
In health care, ambient documentation tools have become the star of the show. These “AI scribes” listen to the patient-physician conversation, transcribe the discussion, and then use generative AI to create a first draft clinical note. These solutions remove the laborious work of fully capturing the patient’s story or the physician’s thoughts about the plan. This technology is seen by some as a miracle. Before large language model chat applications were popularized, many physicians did not think such solutions were possible during our careers.
Conflicting visions for AI’s impact
Organizational leaders are stoked about the future AI can potentially create. There are endless administrative inefficiencies that affect patient care or inflate costs.
Frontline clinicians are still quite wary. Their concerns stem from the technology itself and also from what organizational or system leaders will do with the new efficiencies gained from the deployment of such technology. Those next steps – the policy changes and operational actions that follow broad AI scribe implementations — are the critical pieces that will determine its success.
Will organizations simply continue to add more patients and more tasks to physicians’ plates? Will we slot generative AI solutions into existing clinic processes that may not serve clinicians’ ideal workflow? Will we integrate these generative AI tools into EHR systems that do not support physicians’ thought patterns or desired storytelling purposes?
Dan Gardner is the coauthor of "How Big Things Get Done," a book that explores why so many billion-dollar projects, from nuclear power plants to Olympic Games, go wrong and how some manage to succeed. Drawing on data from over 16,000 megaprojects, Gardner and his coauthor Bent Flyvbjerg reveal the startling truth: Only 0.5% of big projects are delivered on time, on budget, and with the promised results. Business Insider interviewed Dan Gardner to learn about some of the world's most high-profile projects, like the Sydney Opera House, which soared 1,400% over budget, and the troubled California High-Speed Rail, which is expected to cost over $100 billion and hasn't moved any passengers yet. He also spotlights the rare successes, like the Empire State Building and the Hoover Dam, to show what’s possible when projects are built on smart planning, strong leadership, and modular thinking.
Of late, Trump tariffs have companies juggling domestic capabilities—technological and otherwise—and reorientating supply chains to support operations in America. But othe
Richard Platt's insight:
Rankings of the reasons why OEMs have already restored. Pinpoints factors that OEMs should prioritize more highly, geopolitical risk and the use of total cost of ownership. Key Report findings:
** Only 26% of manufacturing workers say their company’s technology is even “somewhat advanced”, with an equal number calling it “very” or “somewhat” outdated. ** 20% of employees say they’ve seen colleagues leave due to outdated systems. ** 57% of manufacturing managers and executives cited cost as their biggest barrier to investing in IT modernization and cybersecurity. ** 51% of manufacturing employees believe U.S. factories are falling behind global competitors in technology modernization and automation. ** 75% of U.S. consumers have a preference for U.S.-made goods, one that has increased as a result of global supply-chain disruptions since the COVID era. ** 62% of consumers said other factors, such as quality and price, ultimately matter more in their purchase decisions. ** 91% of consumers said they are concerned about cybersecurity threats to U.S. manufacturers, with 30% saying they are “very” or “extremely” concerned. ** 47% of those expressing concern about cyberattacks pointed to potential threats from foreign countries such as Russia and China as a cause for anxiety.
OpenAI published a 34-page guide to building AI agents, drawing on insights from its customer deployments. It covers opportunity identification, agent design, and best practices for ensuring safe and effective performance.
Breakdown:
Advances in reasoning, multimodality, and tool use have led to LLM-powered agents, systems that independently perform tasks.
OpenAI recommends use cases once resistant to automation due to complex decisions, brittle rules, or heavy reliance on unstructured data.
At its core, an agent has three components: model, tools, and instructions. It requires three types of tools: data, action, and orchestration.
Orchestration: single-agent and multi-agent systems with Manager (agents as tools) and Decentralized (agents handing off to agents) patterns.
Set up guardrails to address identified use case risks, as shown in the diagram above, and add more as new vulnerabilities are discovered.
Why it’s important: Agents mark a new era in automation, where systems can reason through ambiguity, take action across tools, and handle multi step tasks with a high degree of autonomy. This guide offers the foundational knowledge to start delivering enterprise value with agents.
The signs you were waiting for to upgrade your GPU...
Richard Platt's insight:
The word bottleneck is enough to send shivers down the spines of PC gamers. And the worst kind of bottleneck is a GPU bottleneck, even though that's ideally what you want on a gaming PC. Most games are GPU-bound, so you want a situation where none of the other components are holding your GPU back. In some cases, though, a weaker or older GPU can actually bottleneck your PC, stopping you from getting the maximum performance your system is capable of. Here are the signs that you need a new GPU to eliminate a significant bottleneck.
From May 2025, Norah O'Donnell's report on why China's spies are on the rise, and what happens when one gets caught in the U.S. From June 2025, Cecilia Vega’s report on the Americans spying for Cuba in the U.S. From July 2022, Scott Pelley's interview with a former top intelligence official in the Saudi Arabian government, Saad Aljabri, who claims the kingdom's ruler plotted to kill him and has taken his children hostage. From August 2019, Anderson Cooper’s interview with Justice and FBI officials, who reveal how they caught a former CIA officer spying for the Chinese. And from March 2025, Bill Whitaker's investigation into the mysterious swarms of drones that have been spotted in the sky above the United States.
Richard Platt's insight:
From May 2025, Norah O'Donnell's report on why China's spies are on the rise, and what happens when one gets caught in the U.S. From June 2025, Cecilia Vega’s report on the Americans spying for Cuba in the U.S. From July 2022, Scott Pelley's interview with a former top intelligence official in the Saudi Arabian government, Saad Aljabri, who claims the kingdom's ruler plotted to kill him and has taken his children hostage. From August 2019, Anderson Cooper’s interview with Justice and FBI officials, who reveal how they caught a former CIA officer spying for the Chinese. And from March 2025, Bill Whitaker's investigation into the mysterious swarms of drones that have been spotted in the sky above the United States.
Then Pedraza was introduced to Microsoft 365 Copilot Chat, the AI companion that helps with work tasks. A group of AI experts recently trained him on how to write effective prompts to quickly generate personalized activities for the students just by typing a few traits of each. He was amazed by the results.
Then Pedraza was introduced to Microsoft 365 Copilot Chat, the AI companion that helps with work tasks. A group of AI experts recently trained him on how to write effective prompts to quickly generate personalized activities for the students just by typing a few traits of each. He was amazed by the results. From skepticism to success: How AI is helping teachers transform classrooms in Peru - Very positive report as you would expect from Microsoft: https://news.microsoft.com/source/latam/features/ai/world-bank-peru-teachers-copilot/?lang=en
China leads in AI innovation in 2025 with DeepSeek V3, challenging the US in global AI dominance. Learn how cost drops and security risks shape this race. China's DeepSeek V3 0324 has become a top-performing non-reasoning model globally, highlighting the country's growing dominance in open-weight AI development. Chinese AI models are often optimized for speed and cost efficiency and are specially used for large-scale deployment. The report showcased that the rise of Chinese AI is a significant milestone and showed how Chinese AI labs are bridging the gap and surpassing their US counterparts in the major area of AI innovation. The rise of Chinese AI has certainly pushed the US back. However, US-based labs like OpenAI, Google, and xAI continue to lead in reasoning models, essential for more complex tasks involving step-by-step problem-solving. OpenAI's o4-mini (high) currently tops the global intelligence index, but competitors like Chinese AI are quickly narrowing the performance gap and leading the AI race. If the Chinese AI model continues to grow, it will soon overthrow the US hegemony in AI innovation and become the leading open-source network in the world.
To get content containing either thought or leadership enter:
To get content containing both thought and leadership enter:
To get content containing the expression thought leadership enter:
You can enter several keywords and you can refine them whenever you want. Our suggestion engine uses more signals but entering a few keywords here will rapidly give you great content to curate.
Scoop.it!
Your new post is loading...
Yes, Chinese foundries can knock out less sophisticated chips with fewer transistors. Through techniques such as double and quadruple patterning, they might even be able to produce ICs with transistor measurements of just 7 nm's. SMIC, China's best-known foundry, appeared to have made one for the Mate 60 Pro, a Huawei smartphone released in 2023. In layman's terms, EUV works by firing a concentrated beam of light off the smoothest mirrors in the world and onto a silicon wafer, where it etches complex circuit designs like an Antman scribe. It uses a wavelength measuring just 13.5 nm, compared with the 193 nm of deep ultraviolet lithography (DUV), its predecessor. The difference is like that between a fat marker and a thin ballpoint to a fine artist. According to various reports China's Huawei, develped an EUV system called laser-induced discharge plasma (LDP) technology that's been going through tests at a Huawei facility in Dongguan. One report says it has been able to generate the 13.5-nanometer wavelength by "vaporizing tin between electrodes and converting it to plasma via high-voltage discharge, where electron-ion collisions produce the required wavelength." "It's a pretty cool technique because it's actually simpler than what ASML does," said Earl Lum, a IC expert at EJL Wireless Research. "It could be cheaper to make the machine because of the strategy that ASML had to use." Trials do not mean a Chinese flavor of EUV is close to commercial deployment, of course, and a few press reports that leave many questions unanswered must be treated with a generous dose of skepticism. ASML's share price fell 7% on March 10, days after the reports about China's apparent EUV breakthrough. But this may have been linked to more general concerns about tariffs and their economic impact. Other chip stocks also suffered.