SoylentNews

chromas writes:

Here's an interesting story someone dropped in IRC:

The radical 1960s schools experiment that created a whole new alphabet – and left thousands of children unable to spell (and yes, I tweaked the sub title to fit into SN's tiny title limit):

The Initial Teaching Alphabet was a radical, little-known educational experiment trialled in British schools (and in other English-speaking countries) during the 1960s and 70s. Billed as a way to help children learn to read faster by making spelling more phonetically intuitive, it radically rewrote the rules of literacy for tens of thousands of children seemingly overnight. And then it vanished without explanation. Barely documented, rarely acknowledged, and quietly abandoned – but never quite forgotten by those it touched.

Why was it only implemented in certain schools – or even, in some cases, only certain classes in those schools? How did it appear to disappear without record or reckoning? Are there others like my mum, still aggrieved by ITA? And what happens to a generation taught to read and write using a system that no longer exists?

[...] Unlike Spanish or Welsh, where letters have consistent sound values, English is a patchwork of linguistic inheritances. Its roughly 44 phonemes – the distinct sounds that make up speech – can each be spelt multiple ways. The long "i" sound alone, as in "eye", has more than 20 possible spellings. And many letter combinations contradict one another across different words: think of "through", "though" and "thought".

It was precisely this inconsistency that Conservative MP Sir James Pitman – grandson of Sir Isaac Pitman, the inventor of shorthand – identified as the single greatest obstacle for young readers. In a 1953 parliamentary debate, he argued that it is our "illogical and ridiculous spelling" which is the "chief handicap" that leads many children to stumble with reading, with lasting consequences for their education. His proposed solution, launched six years later, was radical: to completely reimagine the alphabet.

The result was ITA: 44 characters, each representing a distinct sound, designed to bypass the chaos of traditional English and teach children to read, and fast. Among the host of strange new letters were a backwards "z", an "n" with a "g" inside, a backwards "t" conjoined with an "h", a bloated "w" with an "o" in the middle. Sentences in ITA were all written in lower case.

[...] The issue isn't simply whether or not ITA worked – the problem is that no one really knows. For all its scale and ambition, the experiment was never followed by a national longitudinal study. No one tracked whether the children who learned to read with ITA went on to excel, or struggle, as they moved through the education system. There was no formal inquiry into why the scheme was eventually dropped, and no comprehensive lessons-learned document to account for its legacy.

The article has a few stories of ITA students who went on to have poor spelling and bad grades in school from teachers who didn't seem to know about ITA.

Original Submission

Arthur T Knackerbracket has processed the following story:

China's aggressive push to develop a domestic semiconductor industry has largely been successful. The country now has fairly advanced fabs that can produce logic chips using 7nm-class process technologies as well as world-class 3D NAND and DRAM memory devices. However, there are numerous high-profile failures due to missed investments, technical shortcomings, and unsustainable business plans. This has resulted in numerous empty fab shells — zombie fabs — around the country, according to DigiTimes.

As of early 2024, China had 44 wafer semiconductor production facilities, including 25 300-mm fabs, five 200-mm wafers, four 150-mm wafers, and seven inactive ones, according to TrendForce. At the time, 32 additional semiconductor fabrication plans were being constructed in the country as part of the Made in China 2025 initiative, including 24 300-mm fabs and nine 200-mm fabs. Companies like SMIC, HuaHong, Nexchip, CXMT, and Silan planned to start production at 10 new fabs, including nine 300-mm fabs and one 200-mm facility by the end of 2024.

However, while China continues to lead in terms of new fabs coming online, the country also leads in terms of fab shells that never got equipped or put to work, thus becoming zombie fabs. Over the past several years, around a dozen high-profile fab projects, which cost investors between $50 billion and $100 billion, went bust.

Many Chinese semiconductor fab projects failed due to a lack of technical expertise amid overambitious goals: some startups aimed at advanced nodes like 14nm and 7nm without having experienced R&D teams or access to necessary wafer fab equipment. These efforts were often heavily reliant on provincial government funding, with little oversight or industry knowledge, which lead to collapse when finances dried up or scandals emerged. Some fab ventures were plagued by fraud or mismanagement, with executives vanishing or being arrested, sometimes with local officials involved.

To add to problems, U.S. export restrictions since 2019 blocked access of Chinese entities to critical chipmaking equipment required to make chips at 10nm-class nodes and below, effectively halting progress on advanced fabs. In addition, worsening U.S.-China tensions and global market shifts further undercut the viability of many of these projects.

[...] Leading chipmakers, such as Intel, TSMC, Samsung, or SMIC have spent decades developing their production technologies and gain experience in chips on their leading-edge nodes. But Chinese chipmakers Wuhan Hongxin Semiconductor Manufacturing Co. (HSMC) and Quanxin Integrated Circuit Manufacturing (QXIC) attempted to take a shortcut and jump straight to 14nm and, eventually, to 7nm-class nodes by hiring executives and hundreds of engineers from TSMC in 2017 – 2019.

Freeman writes:

https://arstechnica.com/health/2025/07/man-fails-to-take-his-medicine-the-flesh-starts-rotting-off-his-leg/

If you were looking for some motivation to follow your doctor's advice or remember to take your medicine, look no further than this grisly tale.

A 64-year-old man went to the emergency department of Brigham and Women's Hospital in Boston with a painful festering ulcer spreading on his left, very swollen ankle.
[...]
The man told doctors it had all started two years prior, when dark, itchy lesions appeared in the area on his ankle—the doctors noted that there were multiple patches of these lesions on both his legs. But about five months before his visit to the emergency department, one of the lesions on his left ankle had progressed to an ulcer. It was circular, red, tender, and deep. He sought treatment and was prescribed antibiotics, which he took. But they didn't help.
[...]
The ulcer grew. In fact, it seemed as though his leg was caving in as the flesh around it began rotting away. A month before the emergency room visit, the ulcer was a gaping wound that was already turning gray and black at the edges. It was now well into the category of being a chronic ulcer.

In a Clinical Problem-Solving article published in the New England Journal of Medicine this week, doctors laid out what they did and thought as they worked to figure out what was causing the man's horrid sore.
[...]
His diabetes was considered "poorly controlled."
[...]
His blood pressure, meanwhile, was 215/100 mm Hg at the emergency department. For reference, readings higher than 130/80 mm Hg on either number are considered the first stage of high blood pressure.
[...]
Given the patient's poorly controlled diabetes, a diabetic ulcer was initially suspected. But the patient didn't have any typical signs of diabetic neuropathy that are linked to ulcers.
[...]
With a bunch of diagnostic dead ends piling up, the doctors broadened their view of possibilities, newly considering cancers, rare inflammatory conditions, and less common conditions affecting small blood vessels (as the MRI has shown the larger vessels were normal). This led them to the possibility of a Martorell's ulcer.

upstart writes:

The tech's mistakes are dangerous, but its potential for abuse when working as intended is even scarier:

Juan Carlos Lopez-Gomez, despite his U.S. citizenship and Social Security card, was arrested on April 16 on an unfounded suspicion of him being an "unauthorized alien." Immigration and Customs Enforcement kept him in county jail for 30 hours "based on biometric confirmation of his identity"—an obvious mistake of facial recognition technology. Another U.S. citizen, Jensy Machado, was held at gunpoint and handcuffed by ICE agents. He was another victim of mistaken identity after someone else gave his home address on a deportation order. This is the reality of immigration policing in 2025: Arrest first, verify later.

That risk only grows as ICE shreds due process safeguards, citizens and noncitizens alike face growing threats from mistaken identity, and immigration policing agencies increasingly embrace error-prone technology, especially facial recognition. Last month, it was revealed that Customs and Border Protection requested pitches from tech firms to expand their use of an especially error-prone facial recognition technology—the same kind of technology used wrongly to arrest and jail Lopez-Gomez. ICE already has nearly $9 million in contracts with Clearview AI, a facial recognition company with white nationalist ties that was at one point the private facial recognition system most used by federal agencies. When reckless policing is combined with powerful and inaccurate dragnet tools, the result will inevitably be more stories like Lopez-Gomez's and Machado's.

Studies have shown that facial recognition technology is disproportionately likely to misidentify people of color, especially Black women. And with the recent rapid increase of ICE activity, facial recognition risks more and more people arbitrarily being caught in ICE's dragnet without rights to due process to prove their legal standing. Even for American citizens who have "nothing to hide," simply looking like the wrong person can get you jailed or even deported.

While facial recognition's mistakes are dangerous, its potential for abuse when working as intended is even scarier. For example, facial recognition lets Donald Trump use ICE as a more powerful weapon for retribution. The president himself admits he's using immigration enforcement to target people for their political opinions and that he seeks to deport people regardless of citizenship. In the context of a presidential administration that is uncommonly willing to ignore legal procedures and judicial orders, a perfectly accurate facial recognition system could be the most dangerous possibility of all: Federal agents could use facial recognition on photos and footage of protests to identify each of the president's perceived enemies, and they could be arrested and even deported without due process rights.

Colossal's Plans To "De-Extinct" The Giant Moa Are Still Impossible

Arthur T Knackerbracket has processed the following story:

Colossal Biosciences has announced plans to “de-extinct” the New Zealand moa, one of the world’s largest and most iconic extinct birds, but critics say the company’s goals remain scientifically impossible.

The moa was the only known completely wingless bird, lacking even the vestigial wings of birds like emus. There were once nine species of moa in New Zealand, ranging from the turkey-sized bush moa (Anomalopteryx didiformis) to the two biggest species, the South Island giant moa (Dinornis robustus) and North Island giant moa (Dinornis novaezealandiae), which both reached heights of 3.6 metres and weights of 230 kilograms.

It is thought that all moa species were hunted to extinction by the mid-15^th century, following the arrival of Polynesian people, now known as Māori, to New Zealand sometime around 1300.

Colossal has announced that it will work with the Indigenous Ngāi Tahu Research Centre, based at the University of Canterbury in New Zealand, along with film-maker Peter Jackson and Canterbury Museum, which holds the largest collection of moa remains in the world. These remains will play a key role in the project, as Colossal aims to extract DNA to sequence and rebuild the genomes for all nine moa species.

As with Colossal’s other “de-extinction” projects, the work will involve modifying the genomes of animals still living today. Andrew Pask at the University of Melbourne, Australia, who is a scientific adviser to Colossal, says that although the moa’s closest living relatives are the tinamou species from Central and South America, they are comparatively small.

This means the project will probably rely on the much larger Australian emu (Dromaius novaehollandiae). “What emus have is very large embryos, very large eggs,” says Pask. “And that’s one of the things that you definitely need to de-extinct a moa.”

An Anonymous Coward writes:

From '123456' Password Exposed Chats for 64 Million McDonald's Job Applicants

Cybersecurity researchers discovered a vulnerability in McHire, McDonald's chatbot job application platform, that exposed the chats of more than 64 million job applicants across the United States.

The flaw was discovered by security researchers Ian Carroll and Sam Curry, who found that the ChatBot's admin panel utilized a test franchise that was protected by weak credentials of a login name "123456" and a password of "123456".

McHire, powered by Paradox.ai and used by about 90% of McDonald's franchisees, accepts job applications through a chatbot named Olivia. Applicants can submit names, email addresses, phone numbers, home addresses, and availability, and are required to complete a personality test as part of the job application process.

Once logged in, the researchers submitted a job application to the test franchise to see how the process worked.

During this test, they noticed that HTTP requests were sent to an API endpoint at /api/lead/cem-xhr, which used a parameter lead_id, which in their case was 64,185,742.

The researchers found that by incrementing and decrementing the lead_id parameter, they were able to expose the full chat transcripts, session tokens, and personal data of real job applicants that previously applied on McHire.

This type of flaw is called an IDOR (Insecure Direct Object Reference) vulnerability, which is when an application exposes internal object identifiers, such as record numbers, without verifying whether the user is actually authorized to access the data.

"During a cursory security review of a few hours, we identified two serious issues: the McHire administration interface for restaurant owners accepted the default credentials 123456:123456, and an insecure direct object reference (IDOR) on an internal API allowed us to access any contacts and chats we wanted," Carroll explained in a writeup about the flaw.

"Together they allowed us and anyone else with a McHire account and access to any inbox to retrieve the personal data of more than 64 million applicants."

In this case, incrementing or decrementing a lead_id number in a request returned sensitive data belonging to other applicants, as the API failed to check if the user had access to the data.

The issue was reported to Paradox.ai and McDonald's on June 30.

McDonald's acknowledged the report within an hour, and the default admin credentials were disabled soon after.

Original Submission

Arthur T Knackerbracket has processed the following story:

Climate change could pose a threat to the technology industry as copper production is vulnerable to drought, while demand may grow to outstrip supply anyway.

According to a report out today from PricewaterhouseCoopers (PwC), copper mines require a steady water supply to function, and many are situated in places around the world that face a growing risk of severe drought due to shifts in climate.

Copper is almost ubiquitous in IT hardware because of its excellent electrical conductivity, from the tracks on circuit boards to cabling and even the interconnects on microchips. PwC's report focuses just on chips, and claims that nearly a third (32 percent) of global semiconductor production will be reliant on copper supplies that are at risk from climate disruption by 2035.

If something is not done to rein in climate change, like drastically cutting greenhouse gas emissions, then the share of copper supply at risk rises to 58 percent by 2050, PwC claims. As this seems increasingly unlikely, it advises both copper exporters and semiconductor buyers to adapt their supply chains and practices if they are to ride out the risk.

Currently, of the countries or territories that supply the semiconductor industry with copper, the report states that only Chile faces severe drought risks. But within a decade, copper mines in the majority of the 17 countries that source the metal will be facing severe drought risks.

PwC says there is an urgent need to strengthen supply chain resilience. Some businesses are taking action, but many investors believe companies should step up their efforts when it comes to de-risking their supply chain, the firm adds.

According to the report, mining companies can alleviate some of the supply issues by investing in desalination plants, improving water efficiency and recycling water.

Semiconductor makers could use alternative materials, diversify their suppliers, and adopt measures such as recycling and taking advantage of the circular economy.

[...] This was backed up recently by the International Energy Agency (IEA), which reckons supplies of copper will fall 30 percent short of the volume required by 2035 if nothing is done to open up new sources.

One solution is for developed countries to do more refining of copper – plus other key metals needed for industry – and form partnerships with developing countries to help open up supplies, executive director Fatih Birol told The Guardian.

Original Submission

Arthur T Knackerbracket has processed the following story:

Humans come from Africa. This wasn’t always obvious, but today it seems as close to certain as anything about our origins.

There are two senses in which this is true. The oldest known hominins, creatures more closely related to us than to great apes, are all from Africa, going back as far as 7 million years ago. And the oldest known examples of our species, Homo sapiens, are also from Africa.

It’s the second story I’m focusing on here, the origin of modern humans in Africa and their subsequent expansion all around the world. With the advent of DNA sequencing in the second half of the 20th century, it became possible to compare the DNA of people from different populations. This revealed that African peoples have the most variety in their genomes, while all non-African peoples are relatively similar at the genetic level (no matter how superficially different we might appear in terms of skin colour and so forth).

In genetic terms, this is what we might call a dead giveaway. It tells us that Africa was our homeland and that it was populated by a diverse group of people – and that everyone who isn’t African is descended from a small subset of the peoples, who left this homeland to wander the globe. Geneticists were confident about this as early as 1995, and the evidence has only accumulated since.

And yet, the physical archaeology and the genetics don’t match – at least, not on the face of it.

Genetics tells us that all living non-African peoples are descended from a small group that left the continent around 50,000 years ago. Barring some wobbles about the exact date, that has been clear for two decades. But archaeologists can point to a great many instances of modern humans living outside Africa much earlier than that.

What is going on? Is our wealth of genetic data somehow misleading us? Or is it true that we are all descended from that last big migration – and the older bones represent populations that didn’t survive?

Eleanor Scerri at the Max Planck Institute of Geoanthropology in Germany and her colleagues have tried to find an explanation.

Upstart has processed the PerfektBlue Bluetooth Vulnerabilities Expose Millions of Vehicles to Remote Code Execution the following story:

Cybersecurity researchers have discovered a set of four security flaws in OpenSynergy's BlueSDK Bluetooth stack that, if successfully exploited, could allow remote code execution on millions of transport vehicles from different vendors.

The vulnerabilities, dubbed PerfektBlue, can be fashioned together as an exploit chain to run arbitrary code on cars from at least three major automakers, Mercedes-Benz, Volkswagen, and Skoda, according to PCA Cyber Security (formerly PCAutomotive). Outside of these three, a fourth unnamed original equipment manufacturer (OEM) has been confirmed to be affected as well.

"PerfektBlue exploitation attack is a set of critical memory corruption and logical vulnerabilities found in OpenSynergy BlueSDK Bluetooth stack that can be chained together to obtain Remote Code Execution (RCE)," the cybersecurity company said.

While infotainment systems are often seen as isolated from critical vehicle controls, in practice, this separation depends heavily on how each automaker designs internal network segmentation. In some cases, weak isolation allows attackers to use IVI access as a springboard into more sensitive zones—especially if the system lacks gateway-level enforcement or secure communication protocols.

The only requirement to pull off the attack is that the bad actor needs to be within range and be able to pair their setup with the target vehicle's infotainment system over Bluetooth. It essentially amounts to a one-click attack to trigger over-the-air exploitation.

"However, this limitation is implementation-specific due to the framework nature of BlueSDK," PCA Cyber Security added. "Thus, the pairing process might look different between various devices: limited/unlimited number of pairing requests, presence/absence of user interaction, or pairing might be disabled completely."

Arthur T Knackerbracket has processed the following story:

Intel Ceo Says It's "Too Late" For Them To Catch Up With Ai Competition -Reportedly Claims Intel Has Fallen Out Of The "Top 10 Semiconductor Companies" As The Firm Lays Off Thousands Across The World

Dark days ahead, or perhaps already here.

Intel has been in a dire state these past few years, with seemingly nothing going right. Its attempt to modernize x86 with a hybrid big.LITTLE architecture, à la ARM, failed to make a meaningful impact in terms of market share gains, only made worse by last-gen's Arrow Lake chips barely registering a response against AMD’s lineup. On the GPU front, the Blue Team served an undercooked product far too late that, while not entirely hopeless, was nowhere near enough to challenge the industry’s dominant players. All of this compounds into a grim reality, seemingly confirmed by new CEO Lip-Bu Tan in a leaked internal conversation today.

According to OregonTech, it's borderline a fight for survival for the once-great American innovation powerhouse as it struggles to even acknowledge being among the top contenders anymore. Despite Tan's insistence, Intel would still rank fairly well given its extensive legacy. While companies like AMD, Nvidia, Apple, TSMC, and even Samsung might be more successful today, smaller chipmakers like Broadcom, MediaTek, Micron, and SK Hynix are not above the Blue Team in terms of sheer impact. Regardless, talking to employees around the world in a Q&A session, Intel's CEO allegedly shared these bleak words: "Twenty, 30 years ago, we were really the leader. Now I think the world has changed. We are not in the top 10 semiconductor companies."

As evident from the quote, this is a far cry from a few decades ago when Intel essentially held a monopoly over the CPU market, making barely perceptible upgrades each generation in order to sustain its dominance. At one time, Intel was so powerful that it considered acquiring Nvidia for $20 billion. The GPU maker is now worth $4 trillion.

upstart writes:

Derinkuyu: A Subterranean Marvel of Ancient Engineering:

Beneath the sun-drenched plains of Cappadocia, where otherworldly "fairy chimney" rock formations pierce the sky, lies a secret world carved into the very heart of the earth.Forget the grand pyramids or towering ziggurats; we're about to descend into Derinkuyu, an ancient metropolis swallowed by the ground, a testament to human resilience and a whisper from a forgotten past.

Imagine a civilization, facing threats we can only dimly perceive, choosing not to build up, but to delve down, creating a labyrinthine sanctuary that could shelter thousands. This isn't just an archaeological site; it's a subterranean saga etched in stone, waiting to unfold its mysteries. Its origins are somewhat debated, but the prevailing archaeological consensus points to construction likely beginning in the Phrygian period (around the 8th-7th centuries BCE). The Phrygians, an Indo-European people who established a significant kingdom in Anatolia, were known for their rock-cut architecture, and Derinkuyu bears hallmarks of their early techniques.

However, the city's expansion and more complex features likely developed over centuries, with significant contributions from later periods, particularly the Byzantine era (roughly 4th to 15th centuries CE). During this time, Cappadocia was a crucial region for early Christianity, and the need for refuge from various invasions and raids, first from Arab forces and later from the Seljuk Turks, would have spurred the further development and extensive use of these underground complexes.

The city served as a refuge during times of conflict, allowing people to escape from invaders. The underground city could accommodate up to 20,000 people, along with their livestock and supplies, making it a significant shelter during turbulent times. The city extends approximately 60 meters deep and consists of multiple levels—around 18 floors! Each level was designed for specific purposes, such as living quarters, storage rooms, and even places of worship.

The geological context is crucial here. Cappadocia's unique landscape is characterized by soft volcanic tuff, formed by ancient eruptions. This malleable rock was relatively easy to carve, yet strong enough to support the extensive network of tunnels and chambers without collapsing – a testament to the engineering acumen of its builders.

Now, let's talk about the ingenuity of the design. Derinkuyu wasn't just a series of haphazard tunnels; it was a carefully planned multi-level settlement designed for extended habitation. Key features include:

• Ventilation Systems: Remarkably, the city possessed sophisticated ventilation shafts that extended down through multiple levels, ensuring a constant supply of fresh air. Some of these shafts are believed to have also served as wells.

• Water Management: Evidence of wells and water storage areas highlights the critical need for a sustainable water supply during times of siege.

• Defensive Measures: The massive circular stone doors, capable of sealing off corridors from the inside, are a clear indication of the city's primary function as a refuge. These "rolling stones" could weigh several tons and were designed to be moved by a small number of people from within.

• Living and Communal Spaces: Excavations have revealed evidence of domestic areas, including kitchens (with soot-stained ceilings indicating fireplaces), sleeping quarters, and communal gathering spaces.

• Agricultural Infrastructure: The presence of stables suggests that livestock were also brought underground, a vital consideration for long-term survival. Storage rooms for grains and other foodstuffs further underscore the self-sufficiency of the city during times of crisis.

• Religious and Educational Facilities: Some levels appear to have housed areas that may have served as chapels or even rudimentary schools, indicating that life, even in hiding, continued beyond mere survival.

An Anonymous Coward writes:

https://www.bleepingcomputer.com/news/security/new-android-taptrap-attack-fools-users-with-invisible-ui-trick/

A novel tap-jacking technique can exploit user interface animations to bypass Android's permission system and allow access to sensitive data or trick users into performing destructive actions, such as wiping the device.

Unlike traditional, overlay-based tap-jacking, TapTrap attacks work even with zero-permission apps to launch a harmless transparent activity on top of a malicious one, a behavior that remains unmitigated in Android 15 and 16.

TapTrap was developed by a team of security researchers at TU Wien and the University of Bayreuth (Philipp Beer, Marco Squarcina, Sebastian Roth, Martina Lindorfer), and will be presented next month at the USENIX Security Symposium.

However, the team has already published a technical paper that outlines the attack and a website that summarizes most of the details.
How TapTrap works

TapTrap abuses the way Android handles activity transitions with custom animations to create a visual mismatch between what the user sees and what the device actually registers.

A malicious app installed on the target device launches a sensitive system screen (permission prompt, system setting, etc.) from another app using 'startActivity()' with a custom low-opacity animation.

"The key to TapTrap is using an animation that renders the target activity nearly invisible," the researchers say on a website that explains the attack.

"This can be achieved by defining a custom animation with both the starting and ending opacity (alpha) set to a low value, such as 0.01," thus making the malicious or risky activity almost completely transparent.

"Optionally, a scale animation can be applied to zoom into a specific UI element (e.g., a permission button), making it occupy the full screen and increasing the chance the user will tap it."

Although the launched prompt receives all touch events, all the user sees is the underlying app that displays its own UI elements, as on top of it is the transparent screen the user actually engages with.

Thinking they interact with the benign app, a user may tap on specific screen positions that correspond to risky actions, such as an "Allow" or "Authorize" buttons on nearly invisible prompts.

Freeman writes:

https://arstechnica.com/ai/2025/07/agi-may-be-impossible-to-define-and-thats-a-multibillion-dollar-problem/

When is an AI system intelligent enough to be called artificial general intelligence (AGI)? According to one definition reportedly agreed upon by Microsoft and OpenAI, the answer lies in economics: When AI generates $100 billion in profits. This arbitrary profit-based benchmark for AGI perfectly captures the definitional chaos plaguing the AI industry.

In fact, it may be impossible to create a universal definition of AGI, but few people with money on the line will admit it.

Over this past year, several high-profile people in the tech industry have been heralding the seemingly imminent arrival of "AGI" (i.e., within the next two years). [...] As Google DeepMind wrote in a paper on the topic: If you ask 100 AI experts to define AGI, you'll get "100 related but different definitions." [...] When companies claim they're on the verge of AGI, what exactly are they claiming?

This isn't just academic navel-gazing. The definition problem has real consequences for how we develop, regulate, and think about AI systems. When companies claim they're on the verge of AGI, what exactly are they claiming?

I tend to define AGI in a traditional way that hearkens back to the "general" part of its name: An AI model that can widely generalize—applying concepts to novel scenarios—and match the versatile human capability to perform unfamiliar tasks across many domains without needing to be specifically trained for them.

However, this definition immediately runs into thorny questions about what exactly constitutes "human-level" performance. Expert-level humans? Average humans? And across which tasks—should an AGI be able to perform surgery, write poetry, fix a car engine, and prove mathematical theorems, all at the level of human specialists? (Which human can do all that?) More fundamentally, the focus on human parity is itself an assumption; it's worth asking why mimicking human intelligence is the necessary yardstick at all.

The latest example of trouble resulting from this definitional confusion comes from the deteriorating relationship between Microsoft and OpenAI. According to The Wall Street Journal, the two companies are now locked in acrimonious negotiations partly because they can't agree on what AGI even means—despite having baked the term into a contract worth over $13 billion.

fliptop writes:

For the first time ever, a company has achieved a market capitalization of $4 trillion. And that company is none other than Nvidia:

The chipmaker's shares rose as much as 2.5% on Wednesday, pushing past the previous market value record ($3.9 trillion), set by Apple in December 2024. Shares in the AI giant later closed at $162.88, shrinking the company's market value to $3.97 trillion.

Nvidia has rallied by more than 70% from its April 4 low, when global stock markets were sent reeling by President Donald Trump's global tariff rollout.

[...] The record value comes as tech giants such as OpenAI, Amazon and Microsoft are spending hundreds of billions of dollars in the race to build massive data centers to fuel the artificial intelligence revolution. All of those companies are using Nvidia chips to power their services, though some are also developing their own.

In the first quarter of 2025 alone, the company reported its revenue soared about 70%, to more than $44 billion. Nvidia said it expects another $45 billion worth of sales in the current quarter.

Also at: ZeroHedge, CNN and AP.

Related: Nvidia Reportedly Raises GPU Prices by 10-15% as Tariffs and TSMC Price Hikes Filter Down

Original Submission

upstart writes:

Apple just released an interesting coding language model - 9to5Mac:

Apple quietly dropped a new AI model on Hugging Face with an interesting twist. Instead of writing code like traditional LLMs generate text (left to right, top to bottom), it can also write out of order, and improve multiple chunks at once.

The result is faster code generation, at a performance that rivals top open-source coding models. Here's how it works.

The nerdy bits

Here are some (overly simplified, in the name of efficiency) concepts that are important to understand before we can move on.

Autoregression

Traditionally, most LLMs have been autoregressive. This means that when you ask them something, they process your entire question, predict the first token of the answer, reprocess the entire question with the first token, predict the second token, and so on. This makes them generate text like most of us read: left to right, top to bottom.

Temperature

LLMs have a setting called temperature that controls how random the output can be. When predicting the next token, the model assigns probabilities to all possible options. A lower temperature makes it more likely to choose the most probable token, while a higher temperature gives it more freedom to pick less likely ones.

Diffusion

An alternative to autoregressive models is diffusion models, which have been more often used by image models like Stable Diffusion. In a nutshell, the model starts with a fuzzy, noisy image, and it iteratively removes the noise while keeping the user request in mind, steering it towards something that looks more and more like what the user requested.