Leading AI companies including Anthropic, Google, OpenAI and Elon Musk's xAI are discovering significant inconsistencies in how their AI reasoning models operate, according to company researchers. The companies have deployed "chain-of-thought" techniques that ask AI models to solve problems step-by-step while showing their reasoning process, but are finding examples of "misbehaviour" where chatbots provide final responses that contradict their displayed reasoning.

METR, a non-profit research group, identified an instance where Anthropic's Claude chatbot disagreed with a coding technique in its chain-of-thought but ultimately recommended it as "elegant." OpenAI research found that when models were trained to hide unwanted thoughts, they would conceal misbehaviour from users while continuing problematic actions, such as cheating on software engineering tests by accessing forbidden databases.

How AI will reshape the future of work? The Washington Post looks at India's $280 billion call-center and "business process outsourcing" industry, which employs over 3 million people.

2023 saw the arrival of a real-time "accent-altering software" — now used by at least 42,000 call center agents: Those who use the software are engaging in "digital whitewashing," critics say, which helps explain why the industry prefers the term "accent translation" over "accent neutralization." But companies say it's delivering results: happier customers, satisfied agents, faster calls.

Many are not convinced. Whatever short-term gains automation may offer to workers, they say, it will ultimately eliminate far more jobs than it creates. They point to the quality assurance process: When callers hear, "this call may be monitored," that now usually refers to an AI system, not a human [which now can review all calls for compliance and tone]... "AI is going to crush entry-level white-collar hiring over the next 24 to 36 months," said Mark Serdar, who has spent his career helping Fortune 500 companies expand their global workforce. "And it's happening faster than most people realize...." Already, chatbots, or "virtual agents," are handling basic tasks like password resets or balance updates. AI systems are writing code, translating emails, onboarding patients, and analyzing applications for credit cards, mortgages and insurance. The human jobs are changing, too. AI "co-pilots" are providing call center agents with instant answers and suggested scripts. At some companies, bots have started handling the calls.

There is no shortage of ominous predictions about the implications for India's labor force. Within a year, there will only be a "minimal" need for call centers, K Krithivasan, CEO of Indian IT company Tata Consultancy Services, recently told the Financial Times. The Brookings Institution found 86 percent of customer service tasks have "high automation potential." More than a quarter of jobs in India have "high exposure" to AI, the International Monetary Fund has warned. "There is a rapid wave coming," said Pratyush Kumar, co-founder of Sarvam, a leading Indian AI firm, which recently helped a major insurance provider make 40 million automated phone calls informing enrollees that their insurance program was expiring. He said corporate clients are all asking him to help reduce headcount...

While AI may be phasing out certain jobs, its defenders say it is also creating different kinds of opportunities. Teleperformance, along with hundreds of other companies, has hired thousands of data annotators in India — many of them women in small towns and rural areas — to label training images and videos for AI systems. Prompt engineers, data scientists, AI trainers and speech scientists are all newly in demand... At some firms, those who previously worked in quality assurance have transitioned to performance coaching, said [Sharath Narayana, co-founder of AI speech tools company Sanas], whose previous firm, Observe.ai, also built QA software. Still, he admits, 10 to 20 percent of workers he observed "could not upskill at all" and were probably let go.

Even the most hopeful admit that workers who can't adapt will fall behind. "It's like the industrial revolution," said Prithvijit Roy, Accenture's former lead for its Global AI Hub. "Some will suffer."
The article also notes that while Indian universities produce over a million engineering graduates each year, "placement rates are falling at leading IT firms; salaries have stagnated."

As recently as 2021, GM "all but eliminated" hybrids from its future product plans, reports the New York Times. "But then a funny thing happened." Car shoppers balked at the high prices of fully electric models and the challenges of charging them. In the last few years, sales of electric vehicles have grown at a much slower rate than automakers once expected. And hybrids have stepped in to fill the gap, accounting for a large and growing share of new car sales... In the first three months of this year, hybrids — including cars that can and cannot be plugged in — made up about 14 percent of all light vehicles sold in the United States, according to the Department of Energy. That was around twice the market share of fully electric vehicles in that period...

Several automakers are slowing the introduction of new electric vehicles, and have accelerated development of new hybrids.
Robb Report looks at the current status of hybrids — and a possible future: "The charging infrastructure in most countries is not yet mature enough to support convenient mass adoption of battery-electric vehicles, and in some territories never will be," says Jonathan Hall, head of research and advanced engineering at U.K.-based consulting group Mahle Powertrain....

Porsche, active in this space since 2010, just hybridized its iconic 911 for this model year. Lamborghini also joined the trend with the debut of its 1,000 hp Revuelto hybrid in 2023. "The company doesn't plan to give up the internal-combustion engine anytime soon," says CTO Rouven Mohr. "We are also considering synthetic fuels to keep ICE vehicles running after 2030."

Hall concurs: "With the emergence of bio-based and even fully synthetic fuels, the link between the ICE and climate change can be broken." Combined with the development of better batteries, this progressive hybrid model could offer the best of both worlds for years to come.

The education-news site EdSurge published "sponsored content" from Minecraft Education this month. "Students light up when they create something meaningful," the article begins. "Self-expression fuels learning, and creativity lies at the heart of the human experience."

But they also argue that "As AI rapidly reshapes software development, computer science education must move beyond syntax drills and algorithmic repetition." Students "must also learn to think systemically..." As AI automates many of the mechanical aspects of programming, the value of CS education is shifting, from writing perfect code to shaping systems, telling stories through logic and designing ethical, human-centered solutions... [I]t's critical to offer computer science experiences that foster invention, expression and design. This isn't just an education issue — it's a workforce one. Creativity now ranks among the top skills employers seek, alongside analytical thinking and AI literacy. As automation reshapes the job market, McKinsey estimates up to 375 million workers may need to change occupations by 2030. The takeaway? We need more adaptable, creative thinkers.

Creative coding, where programming becomes a medium for self-expression and innovation, offers a promising solution to this disconnect. By positioning code as a creative tool, educators can tap into students' intrinsic motivation while simultaneously building computational thinking skills. This approach helps students see themselves as creators, not just consumers, of technology. It aligns with digital literacy frameworks that emphasize critical evaluation, meaningful contribution and not just technical skills.

One example of creative coding comes from a curriculum that introduces computer science through game design and storytelling in Minecraft... Developed by Urban Arts in collaboration with Minecraft Education, the program offers middle school teachers professional development, ongoing coaching and a 72-session curriculum built around game-based instruction. Designed for grades 6-8, the project-based program is beginner-friendly; no prior programming experience is required for teachers or students. It blends storytelling, collaborative design and foundational programming skills with a focus on creativity and equity.... Students use Minecraft to build interactive narratives and simulations, developing computational thinking and creative design... Early results are promising: 93 percent of surveyed teachers found the Creative Coders program engaging and effective, noting gains in problem-solving, storytelling and coding, as well as growth in critical thinking, creativity and resilience.

As AI tools like GitHub Copilot become standard in development workflows, the definition of programming proficiency is evolving. Skills like prompt engineering, systems thinking and ethical oversight are rising in importance, precisely what creative coding develops... As AI continues to automate routine tasks, students must be able to guide systems, understand logic and collaborate with intelligent tools. Creative coding introduces these capabilities in ways that are accessible, culturally relevant and engaging for today's learners.
Some background from long-time Slashdot reader theodp: The Urban Arts and Microsoft Creative Coders program touted by EdSurge in its advertorial was funded by a $4 million Education Innovation and Research grant that was awarded to Urban Arts in 2023 by the U.S. Education Department "to create an engaging, game-based, middle school CS course using Minecraft tools" for 3,450 middle schoolers (6th-8th grades)" in New York and California (Urban Arts credited Minecraft for helping craft the winning proposal)... New York City is a Minecraft Education believer — the Mayor's Office of Media and Entertainment recently kicked off summer with the inaugural NYC Video Game Festival, which included the annual citywide Minecraft Education Battle of the Boroughs Esports Competition in partnership with NYC Public Schools.

In April Anthorpic introduced a new AI trick: multiple Claude agents combine for a "Research" feature that can "search across both your internal work context and the web" (as well as Google Workspace "and any integrations...")

But a recent Anthropic blog post notes this feature "involves an agent that plans a research process based on user queries, and then uses tools to create parallel agents that search for information simultaneously," which brings challenges "in agent coordination, evaluation, and reliability.... The model must operate autonomously for many turns, making decisions about which directions to pursue based on intermediate findings." Multi-agent systems work mainly because they help spend enough tokens to solve the problem.... This finding validates our architecture that distributes work across agents with separate context windows to add more capacity for parallel reasoning. The latest Claude models act as large efficiency multipliers on token use, as upgrading to Claude Sonnet 4 is a larger performance gain than doubling the token budget on Claude Sonnet 3.7. Multi-agent architectures effectively scale token usage for tasks that exceed the limits of single agents.

There is a downside: in practice, these architectures burn through tokens fast. In our data, agents typically use about 4Ã-- more tokens than chat interactions, and multi-agent systems use about 15Ã-- more tokens than chats. For economic viability, multi-agent systems require tasks where the value of the task is high enough to pay for the increased performance. Further, some domains that require all agents to share the same context or involve many dependencies between agents are not a good fit for multi-agent systems today.

For instance, most coding tasks involve fewer truly parallelizable tasks than research, and LLM agents are not yet great at coordinating and delegating to other agents in real time. We've found that multi-agent systems excel at valuable tasks that involve heavy parallelization, information that exceeds single context windows, and interfacing with numerous complex tools.
Thanks to Slashdot reader ZipNada for sharing the news.

Longtime Slashdot reader fahrbot-bot shares a report from the Cool Down: A team of chemical engineers at the Massachusetts Institute of Technology has invented a new process to separate crude oil components, potentially bringing forward a replacement that can cut its harmful carbon pollution by 90%. The original technique, which uses heat to separate crude oil into gasoline, diesel, and heating oil, accounts for roughly 1% of all global energy consumption and 6% of dirty energy pollution from the carbon dioxide it releases.

"Instead of boiling mixtures to purify them, why not separate components based on shape and size?" said Zachary P. Smith, associate professor of chemical engineering at MIT and senior author of the study, as previously reported in Interesting Engineering. The team invented a polymer membrane that divides crude oil into its various uses like a sieve. The new process follows a similar strategy used by the water industry for desalination, which uses reverse osmosis membranes and has been around since the 1970s. [The membrane excelled in lab tests. It increased the toluene concentration by 20 times in a mixture with triisopropylbenzene. It also effectively separated real industrial oil samples containing naphtha, kerosene, and diesel.]

Hackers have stolen hundreds of millions of dollars from cryptocurrency holders and disrupted major retailers by targeting outsourced call centers used by American corporations to reduce costs, WSJ reported Thursday. The attackers exploit low-paid call center workers through bribes and social engineering to bypass two-factor authentication systems protecting bank accounts and online portals.

Coinbase faces potential losses of $400 million after hackers compromised data belonging to 97,000 customers by bribing call center workers in India with payments of $2,500. The criminals also used malicious tools that exploited vulnerabilities in Chrome browser extensions to collect customer data in bulk.

TaskUs, which handled Coinbase support calls, shut down operations at its Indore, India facility and laid off 226 workers. Retail attacks targeted Marks & Spencer and Harrods with hackers impersonating corporate executives to pressure tech support workers into providing network access. The same technique compromised MGM Resorts systems in 2023. Call center employees typically possess sensitive customer information including account balances and recent transactions that criminals use to masquerade as legitimate company representatives.

Microsoft is planning to ax thousands of jobs, particularly in sales, as part of the company's latest move to trim its workforce amid heavy spending on AI. From a report: The cuts are expected to be announced early next month [non-paywalled source], following the end of Microsoft's fiscal year, according to people familiar with the matter. The reductions won't exclusively affect sales teams, and the timing could still change, said the people, who requested anonymity to discuss a private matter. The terminations would follow a previous round of layoffs in May that hit 6,000 people and fell hardest on product and engineering positions, largely sparing customer-facing roles like sales and marketing.

Longtime Slashdot reader UnknowingFool writes: A new documentary released last week on Netflix goes into detail about events leading up to the destruction of OceanGate's submersible, Titan that imploded on June 18, 2023 while attempting to visit the wreckage of the RMS Titanic off the coast of Newfoundland. The Titan used a carbon-fiber hull instead of more traditional materials like steel or titanium. "Through exclusive access to whistleblower testimony, pivotal audio recordings, and footage from the company's early days, the film provides an unprecedented look at the technical challenges, moral dilemmas, and shockingly poor decisions that culminated in the catastrophic expedition," explains Netflix in an article.

Some highlights:
- Titan's original carbon-fiber hull had been replaced with a second carbon-fiber one after the first one developed noticeable cracks.
- Three scale models of the second hull failed tests. OceanGate decided to manufacture the second hull regardless of these failures.
- Loud pops were heard in many dives; CEO Stockton Rush dismissed these as "seasoning".
- Many employees raised numerous safety concerns. They were fired like lead pilot and head of marine operations, David Lochridge. Or they quit.
- Some employees like Emily Hammermeister wanted to quit earlier, but external conditions like the COVID pandemic made it difficult. After the scale models failed, she refused to bolt anyone in the future submersible. She was given the two options of being fired or quit; she quit in the middle of the pandemic.
- Rush's blindness to inconvenient facts: After the crack was discovered, Rush questioned Director of Engineering, Tony Nissen, about why Nissen did not anticipate the possibility of a crack. Nissen: "I wrote you a report that showed you it was there." Nissen had warned repeatedly that the hull's fibers were breaking (the pops) with each dive. Rush: "Well, one of us has to go."
- Poor decisions by Rush extended beyond engineering decisions. After Rush fired Lochridge for raising safety concerns , Rush wanted Bonnie Carl, the company's accountant, to be his replacement pilot. While Carl was an experienced scuba diver, she quit as she was extremely uncomfortable being a pilot. Her explanation: "Are you nuts? I'm an accountant."

Google Cloud has attributed last week's widespread outage to a flawed code update in its Service Control system that triggered a global crash loop due to missing error handling and lack of feature flag protection. The Register reports: Google's explanation of the incident opens by informing readers that its APIs, and Google Cloud's, are served through our Google API management and control planes." Those two planes are distributed regionally and "are responsible for ensuring each API request that comes in is authorized, has the policy and appropriate checks (like quota) to meet their endpoints." The core binary that is part of this policy check system is known as "Service Control."

On May 29, Google added a new feature to Service Control, to enable "additional quota policy checks." "This code change and binary release went through our region by region rollout, but the code path that failed was never exercised during this rollout due to needing a policy change that would trigger the code," Google's incident report explains. The search monopolist appears to have had concerns about this change as it "came with a red-button to turn off that particular policy serving path." But the change "did not have appropriate error handling nor was it feature flag protected. Without the appropriate error handling, the null pointer caused the binary to crash."

Google uses feature flags to catch issues in its code. "If this had been flag protected, the issue would have been caught in staging." That unprotected code ran inside Google until June 12th, when the company changed a policy that contained "unintended blank fields." Here's what happened next: "Service Control, then regionally exercised quota checks on policies in each regional datastore. This pulled in blank fields for this respective policy change and exercised the code path that hit the null pointer causing the binaries to go into a crash loop. This occurred globally given each regional deployment."

Google's post states that its Site Reliability Engineering team saw and started triaging the incident within two minutes, identified the root cause within 10 minutes, and was able to commence recovery within 40 minutes. But in some larger Google Cloud regions, "as Service Control tasks restarted, it created a herd effect on the underlying infrastructure it depends on ... overloading the infrastructure." Service Control wasn't built to handle this, which is why it took almost three hours to resolve the issue in its larger regions. The teams running Google products that went down due to this mess then had to perform their own recovery chores. Going forward, Google has promised a couple of operational changes to prevent this mistake from happening again: "We will improve our external communications, both automated and human, so our customers get the information they need asap to react to issues, manage their systems and help their customers. We'll ensure our monitoring and communication infrastructure remains operational to serve customers even when Google Cloud and our primary monitoring products are down, ensuring business continuity."

"Cooper Taylor is only 17 years old, but he's already trying to revolutionize the drone industry," writes Business Insider: His design makes the drone more efficient, customizable, and less expensive to construct, he says. He's built six prototypes, 3D printing every piece of hardware, programming the software, and even soldering the control circuit board. He says building his drone cost one-fifth of the price of buying a comparable machine, which sells for several thousand dollars. Taylor told Business Insider he hopes that "if you're a first responder or a researcher or an everyday problem solver, you can have access to this type of drone."

His innovation won him an $8,000 scholarship in April at the Junior Science and Humanities Symposium, funded by the Defense Department. Then, on May 16, he received an even bigger scholarship of $15,000 from the US Navy, which he won after presenting his research at the Regeneron International Science and Engineering Fair...

It all started when Taylor's little sister got a drone, and he was disappointed to see that it could fly for only about 30 minutes before running out of power. He did some research and found that a vertical take-off and landing, or VTOL, drone would last longer. This type of drone combines the multi-rotor helicopter style with the fixed wings of an airplane, making it extremely versatile. It lifts off as a helicopter, then transitions into plane mode. That way, it can fly farther than rotors alone could take it, which was the drawback to Taylor's sister's drone. Unlike a plane-style drone, though, it doesn't need a runway, and it can hover with its helicopter rotors.
Taylor designed a motor "that could start out helicopter-style for liftoff, then tilt back to become an airplane-style motor," according to the article.

And now this summer he'll be "working on a different drone project through a program with the Reliable Autonomous Systems Lab at the Massachusetts Institute of Technology."

Thanks to Slashdot reader Agnapot for sharing the news.

In a world first, a research team used 2D materials — only an atom thick — to develop a computer. The team (led by researchers at Pennsylvania State University) says it's a major step toward thinner, faster and more energy-efficient electronics.

From the University's announcement: They created a complementary metal-oxide semiconductor (CMOS) computer — technology at the heart of nearly every modern electronic device — without relying on silicon. Instead, they used two different 2D materials to develop both types of transistors needed to control the electric current flow in CMOS computers: molybdenum disulfide for n-type transistors and tungsten diselenide for p-type transistors... "[A]s silicon devices shrink, their performance begins to degrade," [said lead researcher/engineering professor Saptarshi Das]. "Two-dimensional materials, by contrast, maintain their exceptional electronic properties at atomic thickness, offering a promising path forward...."

The team used metal-organic chemical vapor deposition (MOCVD) — a fabrication process that involves vaporizing ingredients, forcing a chemical reaction and depositing the products onto a substrate — to grow large sheets of molybdenum disulfide and tungsten diselenide and fabricate over 1,000 of each type of transistor. By carefully tuning the device fabrication and post-processing steps, they were able to adjust the threshold voltages of both n- and p-type transistors, enabling the construction of fully functional CMOS logic circuits.

"Our 2D CMOS computer operates at low-supply voltages with minimal power consumption and can perform simple logic operations at frequencies up to 25 kilohertz," said first author Subir Ghosh, a doctoral student pursuing a degree in engineering science and mechanics under Das's mentorship. Ghosh noted that the operating frequency is low compared to conventional silicon CMOS circuits, but their computer — known as a one instruction set computer — can still perform simple logic operations.

Edd Gent reporting for IEEE Spectrum: IBM has unveiled a new quantum computing architecture it says will slash the number of qubits required for error correction. The advance will underpin its goal of building a large-scale, fault-tolerant quantum computer, called Starling, that will be available to customers by 2029. Because of the inherent unreliability of the qubits (the quantum equivalent of bits) that quantum computers are built from, error correction will be crucial for building reliable, large-scale devices. Error-correction approaches spread each unit of information across many physical qubits to create "logical qubits." This provides redundancy against errors in individual physical qubits.

One of the most popular approaches is known as a surface code, which requires roughly 1,000 physical qubits to make up one logical qubit. This was the approach IBM focused on initially, but the company eventually realized that creating the hardware to support it was an "engineering pipe dream," Jay Gambetta, the vice president of IBM Quantum, said in a press briefing. Around 2019, the company began to investigate alternatives. In a paper published in Nature last year, IBM researchers outlined a new error-correction scheme called quantum low-density parity check (qLDPC) codes that would require roughly one-tenth of the number of qubits that surface codes need. Now, the company has unveiled a new quantum-computing architecture that can realize this new approach. "We've cracked the code to quantum error correction and it's our plan to build the first large-scale, fault-tolerant quantum computer," said Gambetta, who is also an IBM Fellow. "We feel confident it is now a question of engineering to build these machines, rather than science."

High-voltage electricity cables have become a major constraint throttling the clean energy transition, with manufacturing facilities booked out for years as demand far exceeds supply capacity. The energy transition, trade barriers, and overdue grid upgrades have turbocharged demand for these highly sophisticated cables that connect wind farms, solar installations, and cross-border power networks.

The International Energy Agency estimates that 80 million kilometers of grid infrastructure must be built between now and 2040 to meet clean energy targets -- equivalent to rebuilding the entire existing global grid that took a century to construct, but compressed into just 15 years. Each high-voltage cable requires custom engineering and months-long production in specialized 200-meter towers, with manufacturers reporting that 80-90% of major projects now use high-voltage direct current technology versus traditional alternating current systems.

An anonymous reader quotes a report from Ars Technica: Last week, electronics engineer Lorentio Brodesco announced the completion of a mock-up for nsOne, reportedly the first custom PlayStation 1 motherboard created outside of Sony in the console's 30-year history. The fully functional board accepts original PlayStation 1 chips and fits directly into the original console case, marking a milestone in reverse-engineering for the classic console released in 1994. Brodesco's motherboard isn't an emulator or FPGA-based re-creation -- it's a genuine circuit board designed to work with authentic PlayStation 1 components, including the CPU, GPU, SPU, RAM, oscillators, and voltage regulators. The board represents over a year of reverse-engineering work that began in March 2024 when Brodesco discovered incomplete documentation while repairing a PlayStation 1.

"This isn't an emulator. It's not an FPGA. It's not a modern replica," Brodesco wrote in a Reddit post about the project. "It's a real motherboard, compatible with the original PS1 chips." It's a desirable project for some PS1 enthusiasts because a custom motherboard could allow owners of broken consoles to revive their systems by transplanting original chips from damaged boards onto new, functional ones. With original PS1 motherboards becoming increasingly prone to failure after three decades, replacement boards could extend the lifespan of these classic consoles without resorting to emulation.

The nsOne project -- short for "Not Sony's One" -- uses a hybrid design based on the PU-23 series motherboards found in SCPH-900X PlayStation models but reintroduces the parallel port that Sony had removed from later revisions. Brodesco upgraded the original two-layer PCB design to a four-layer board while maintaining the same form factor. [...] As Brodesco noted on Kickstarter, his project's goal is to "create comprehensive documentation, design files, and production-ready blueprints for manufacturing fully functional motherboards." Beyond repairs, the documentation and design files Brodesco is creating would preserve the PlayStation 1's hardware architecture for future generations: "It's a tribute to the PS1, to retro hardware, and to the belief that one person really can build the impossible."

An anonymous reader quotes a report from The Register: Australian SaaS-y graphic design service Canva now requires candidates for developer jobs to use AI coding assistants during the interview process. [...] Canva's hiring process previously included an interview focused on computer science fundamentals, during which it required candidates to write code using only their actual human brains. The company now expects candidates for frontend, backend, and machine learning engineering roles to demonstrate skill with tools like Copilot, Cursor, and Claude during technical interviews, Canva head of platforms Simon Newton wrote in a Tuesday blog post.

His rationale for the change is that nearly half of Canva's frontend and backend engineers use AI coding assistants daily, that it's now expected behavior, and that the tools are "essential for staying productive and competitive in modern software development." Yet Canva's old interview process "asked candidates to solve coding problems without the very tools they'd use on the job," Newton admitted. "This dismissal of AI tools during the interview process meant we weren't truly evaluating how candidates would perform in their actual role," he added. Candidates were already starting to use AI assistants during interview tasks -- and sometimes used subterfuge to hide it. "Rather than fighting this reality and trying to police AI usage, we made the decision to embrace transparency and work with this new reality," Newton wrote. "This approach gives us a clearer signal about how they'll actually perform when they join our team." The initial reaction among engineers "was worry that we were simply replacing rigorous computer science fundamentals with what one engineer called 'vibe-coding sessions,'" Newton said.

The company addressed these concerns with a recruitment process that sees candidates expected to use their preferred AI tools, to solve what Newton described as "the kind of challenges that require genuine engineering judgment even with AI assistance." Newton added: "These problems can't be solved with a single prompt; they require iterative thinking, requirement clarification, and good decision-making."

Apple executives defended the company's AI strategy this week after acknowledging that major Siri features announced at last year's Worldwide Developers Conference remain undelivered and were quietly pulled from development plans. Craig Federighi, Apple's senior vice president of software engineering, told the Wall Street Journal that the company is rebuilding Siri from the ground up, admitting that while Apple had working software for the promised features, "it didn't converge in the way quality-wise that we needed it to."

The missing capabilities included Siri's ability to search through apps and respond to on-screen activities, features that were demonstrated a year ago but never shipped to users. In the upcoming iOS 26, Apple has instead incorporated more OpenAI technology, allowing users to interact with ChatGPT through camera and screenshots and generate images using OpenAI's tools. Federighi defended the strategy by comparing Apple's position to the early internet era, when the company focused on making other services accessible rather than building competing platforms.

The Class of 2025 is encountering the worst entry-level job market in years with unemployment among recent degree-holders aged 22 to 27 reaching 5.8% this spring -- the highest level in approximately four years and well above the national average. According to Federal Reserve Bank of New York data, 85% of the unemployment rate increase since mid-2023 stems from new labor market entrants struggling to find work.

Corporate hiring freezes implemented under threats of President Trump's tariffs, combined with AI replacing traditional entry-level positions, have severely constrained opportunities for new graduates. More than 60% of executives surveyed on LinkedIn indicate that AI will eventually assume tasks currently assigned to entry-level employees, particularly mundane and manual roles.

The impact varies significantly by major, with computer engineering graduates -- once highly sought-after -- now facing a 7.5% unemployment rate, the third-highest among recent graduates. Employment in computer science and mathematical jobs for those under 27 has dropped 8% since 2022, even as it grew 0.8% for older workers.

An anonymous reader quotes a report from TechCrunch: Apple is launching what it calls the Foundation Models framework, which the company says will let developers tap into its AI models in an offline, on-device fashion. Onstage at WWDC 2025 on Monday, Apple VP of software engineering Craig Federighi said that the Foundation Models framework will let apps use on-device AI models created by Apple to drive experiences. These models ship as a part of Apple Intelligence, Apple's family of models that power a number of iOS features and capabilities.

"For example, if you're getting ready for an exam, an app like Kahoot can create a personalized quiz from your notes to make studying more engaging," Federighi said. "And because it happens using on-device models, this happens without cloud API costs [] We couldn't be more excited about how developers can build on Apple intelligence to bring you new experiences that are smart, available when you're offline, and that protect your privacy."

In a blog post, Apple says that the Foundation Models framework has native support for Swift, Apple's programming language for building apps for its various platforms. The company claims developers can access Apple Intelligence models with as few as three lines of code. Guided generation, tool calling, and more are all built into the Foundation Models framework, according to Apple. Automattic is already using the framework in its Day One journaling app, Apple says, while mapping app AllTrails is tapping the framework to recommend different hiking routes.

The New York Times reports: California State University announced this year that it was making ChatGPT available to more than 460,000 students across its 23 campuses to help prepare them for "California's future A.I.-driven economy." Cal State said the effort would help make the school "the nation's first and largest A.I.-empowered university system..." Some faculty members have already built custom chatbots for their students by uploading course materials like their lecture notes, slides, videos and quizzes into ChatGPT.
And other U.S. campuses including the University of Maryland are also "working to make A.I. tools part of students' everyday experiences," according to the article. It's all part of an OpenAI initiative "to overhaul college education — by embedding its artificial intelligence tools in every facet of campus life."

The Times calls it "a national experiment on millions of students." If the company's strategy succeeds, universities would give students A.I. assistants to help guide and tutor them from orientation day through graduation. Professors would provide customized A.I. study bots for each class. Career services would offer recruiter chatbots for students to practice job interviews. And undergrads could turn on a chatbot's voice mode to be quizzed aloud ahead of a test. OpenAI dubs its sales pitch "A.I.-native universities..." To spread chatbots on campuses, OpenAI is selling premium A.I. services to universities for faculty and student use. It is also running marketing campaigns aimed at getting students who have never used chatbots to try ChatGPT...

OpenAI's campus marketing effort comes as unemployment has increased among recent college graduates — particularly in fields like software engineering, where A.I. is now automating some tasks previously done by humans. In hopes of boosting students' career prospects, some universities are racing to provide A.I. tools and training...

[Leah Belsky, OpenAI's vice president of education] said a new "memory" feature, which retains and can refer to previous interactions with a user, would help ChatGPT tailor its responses to students over time and make the A.I. "more valuable as you grow and learn." Privacy experts warn that this kind of tracking feature raises concerns about long-term tech company surveillance. In the same way that many students today convert their school-issued Gmail accounts into personal accounts when they graduate, Ms. Belsky envisions graduating students bringing their A.I. chatbots into their workplaces and using them for life.

"It would be their gateway to learning — and career life thereafter," Ms. Belsky said.