Mathematics is often regarded as the ideal domain for measuring AI progress effectively. Math’s step-by-step logic is easy to track, and its definitive automatically verifiable answers remove any human or subjective factors. But AI systems are improving at such a pace that math benchmarks are struggling to keep up.

Way back in November 2024, non-profit research organization Epoch AI quietly released FrontierMath. A standardized, rigorous benchmark, Frontier Math was designed to measure the mathematical reasoning capabilities of the latest AI tools.

“It’s a bunch of really hard math problems,” explains Greg Burnham, Epoch AI Senior Researcher. “Originally, it was 300 problems that we now call tiers 1–3, but having seen AI capabilities really speed up, there was a feeling that we had to run to stay ahead, so now there’s a special challenge set of extra carefully constructed problems that we call tier 4.”

To a rough approximation, tiers 1–4 go from advanced undergraduate through to early postdoc level mathematics. When introduced, state-of-the-art AI models were unable to solve more than 2% of the problems FrontierMath contained. Fast forward to today and the best publicly available AI models, such as GPT-5.2 and Claude Opus 4.6, are solving over 40% of FrontierMath’s 300 tiers 1–3 problems, and over 30% of the 50 tier 4 problems.

AI takes on PhD level mathematics

And this dizzying pace of advancement is showing no signs of abating. For example, just recently Google DeepMind announced that Aletheia, an experimental AI system derived from Gemini Deep Think, achieved publishable PhD level research results. Though obscure mathematically—calculating certain structure constants in arithmetic geometry called eigenweights—the result is significant in terms of AI development.

“They’re claiming it was essentially autonomous, meaning a human wasn’t guiding the work, and it’s publishable,” Burnham says. “It’s definitely at the lower end of the spectrum of work that would get a mathematician excited, but it’s new—it’s something we truly haven’t really seen before.”

To place this achievement in context, every FrontierMath problem has a known answer that a human has derived. Though a human could probably have achieved Aletheia’s result “if they sat down and steeled themselves for a week,” says Burnham, no human had ever done so.

Aletheia’s results and other recent achievements by AI mathematicians point to new, tougher benchmarks being needed to understand AI capabilities, and fast, because existing ones will soon become irrelevant. “There are easier math benchmarks that are already obsolete, several generations of them,” says Burnham. “FrontierMath will probably saturate [meaning state-of-the-art AI models score 100%] within the next two years; could be faster.”

The First Proof challenge

To begin to address this problem, on February 6, a group of 11 highly distinguished mathematicians proposed the First Proof challenge, a set of 10 extremely difficult math questions which arose naturally in the authors’ research processes, and whose proofs are roughly five pages or less and had not been shared with anyone. The First Proof challenge was a preliminary effort to assess the capabilities of AI systems in solving research-level math questions on their own.

Generating serious buzz in the math community, professional and amateur mathematicians, and teams including OpenAI, all stepped up to the challenge. But by the time the authors posted the proofs on February 14, no one had submitted correct solutions to all 10 problems.

In fact, far from it. The authors themselves only solved two of the 10 problems using Gemini 3.0 Deep Think and ChatGPT 5.2 Pro. And most outside submissions fared little better, apart from OpenAI and a small Aletheia team at Google DeepMind. With “limited human supervision” OpenAI’s most advanced internal AI system solved five of the 10 problems, with Aletheia achieving similar outcomes—results met with a spectrum of emotions by different members of the mathematics community, from awe to disappointment. The team behind First Proof plans an even tougher second round on March 14.

A new frontier for AI

“I think First Proof is terrific: it’s as close as you could realistically get to putting an AI system in the shoes of a mathematician,” says Burnham. Though he admires how First Proof tests AI’s mathematical utility for a wide range of mathematics and mathematicians, Epoch AI has its own new approach to testing—FrontierMath: Open Problems. Uniquely, the pilot benchmark consists of 16 open problems (with more to follow) from research mathematics that professional mathematicians have tried and failed to solve. Since Open Problems’ release on January 27, none have been solved by an AI.

“With Open Problems, we’ve tried to make it more challenging,” says Burnham. “The baseline on its own would be publishable, at least in a specialty journal.” What’s more, each question is designed so that it can be automatically graded. “This is a bit counterintuitive,” Burnham adds. “No one knows the answers, but we have a computer program that will be able to judge whether the answer is right or not.”

Burnham sees First Proof and Open Problems as being complementary. “I would say understanding AI capabilities is a more-the-merrier situation,” he adds. “AI has gotten to the point where it’s, in some ways, better than most PhD students, so we need to pose problems where the answer would be at least moderately interesting to some human mathematicians, not because AI was doing it, but because it’s mathematics that human mathematicians care about.”

Survivor 50 is about to play the biggest game in Fiji history, and reality TV fans don’t have to break the bank to catch this season’s murderers’ row of castaways. Here’s the play:

You can watch Survivor for free in Canada. Traveling abroad in Australia, UK, Asia, India, etc? Theres a handy fix below (TLDR: get this VPN now to watch free anywhere).

The new skilled roles will be created as part of an initial five-year $350m investment.

AI infrastructure provider Equinix will create 200 jobs in Dundalk, Co Louth through an investment of up to $700m in a new facility that will be built by local company Hanley Energy.

New roles at the facility will be in technical fields such as precision engineering, quality assurance and lean manufacturing. Hiring has already begun, and there are plans for the facility to engage locally with apprenticeship and training programmes.

The deal guarantees an initial $350m across five years and is extendable to 10 years. The new facility will be used exclusively to build specialised power equipment for Equinix’s data centres worldwide.

Adaire Fox-Martin, the CEO and president of Equinix, said: “This investment builds upon Equinix’s longtime presence in Ireland and reflects the strategically important role the country plays in the global technology ecosystem.

“Our expansion in Dundalk further strengthens our ability to meet growing customer demand while creating local jobs and supporting the community.”

Equinix said that by centralising its production of components such as low-voltage switchgear, power distribution units and remote power panels at the new 150,000 sq ft facility, it expects to achieve up to 15pc faster lead times when compared to traditional procurement methods.

Taoiseach Micheál Martin, TD said: “This significant announcement reinforces Ireland’s position as a leader in digital infrastructure and advanced manufacturing. The creation of hundreds of skilled jobs and the introduction of world-class facilities in Dundalk is a major boost for the region and for our national economy.”

Equinix, which runs 280 data centres across six continents, works with partners such as Amazon Web Services, Nvidia, Azure and Google Cloud.

Managing director of Equinix Ireland Peter Lantry said the investment would secure the company’s supply chain locally in the Louth region and ensure its long-term presence there, while IDA Ireland CEO Michael Lohan said it would “deliver significant economic benefits and high-value jobs in Dundalk and the wider region”.

Hanley Energy, which employs around 850 worldwide, plans to build the new facility using “low-carbon materials and efficient construction practices”, according to Equinix. The facility will also feature a temperature-controlled testing laboratory – which Equinix claimed will be the only one of its kind in Ireland or the UK.

Hanley CEO John O’Driscoll said: “Partnering with Equinix on this transformative project highlights the strength of Irish engineering and innovation. Our advanced testing facilities and expertise will ensure that the equipment produced here meets the highest global standards, supporting data centres worldwide.”

At the end of last year, Hanley was acquired by US engineering and manufacturing company Jabil for around $725m.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

This article is crossposted from IEEE Spectrum’s careers newsletter. Sign up now to get insider tips, expert advice, and practical strategies, written in partnership with tech career development company Parsity and delivered to your inbox for free!

Standing Out as a Remote Worker Takes a Different Strategy

My first experience as a remote worker was a disaster.

Before I joined a San Francisco-based team with a lead developer in Connecticut, I had worked in person, five days a week. I thought success was simple: write good code, solve hard problems, deliver results. So I put my head down and worked harder than ever.

Twelve-hour days became normal as the boundary between work and personal life disappeared. My kitchen table became my office.

I rarely asked for help because I didn’t want to seem incompetent. I stayed quiet in team Slack channels because I wasn’t sure what to say.

Despite working some of the longest hours of my career, I made the slowest progress. I felt disconnected from the team. I had no idea if my work mattered or if anyone noticed what I was doing. I was burning out.

Eventually, I realized the real problem: I was invisible.

The Office Advantage You Lose When Remote

In an office, visibility happens naturally. Colleagues see you arrive early or stay late. They notice when you are stuck on a problem. They hear about your work in hallway conversations and over lunch. Physical presence creates recognition with almost no effort.

Remote work removes those signals. Your manager cannot see you at your desk. Your teammates don’t know you’ve hit a roadblock unless you say so. You can work long days and still appear less engaged than someone in the office.

That is the shift many people miss: Remote work requires execution plus deliberate communication.

What Actually Works

By my second remote role, I knew I had to change to protect my sanity and still succeed.

Here are five things I did that made a real difference.

1. Over-communicating

I began sharing updates in team channels regularly, not just when asked. “Working on the payment integration today; ready for review tomorrow.” “Hit a blocker with API rate limits; investigating options.” These took seconds but made my work visible and invited help sooner.

2. Setting limits

When your home is also your office, overwork becomes the default. I started ending most days at 5 p.m. and transitioning out of work mode with a walk or gym session. That ritual helped prevent burnout.

3. Volunteering for presentations

Presenting remotely felt less intimidating than standing in front of a room. I started volunteering for demos and lunch-and-learns. This increased my visibility beyond my immediate team and improved my communication skills.

4. Promoting others publicly

When someone helped me, I thanked them in a public channel. When a teammate shipped something impressive, I called it out. This builds goodwill and signals collaboration. In remote environments, gratitude is visible and memorable.

5. Building relationships deliberately

In an office, relationships form naturally. Remotely, you have to create those moments. I started an engineering book club that met every other week to discuss a technical book. It became a low-pressure way to connect with people across the organization.

The Counterintuitive Reality

With these habits, I got promoted faster in this remote job than I ever did in an office. I moved from senior engineer to engineering manager in under two years, while maintaining a better work-life balance.

Remote work offers flexibility and freedom, but it comes with a tax. You are easier to overlook and more likely to burn out unless you are intentional in your actions.

So, succeeding remotely takes deliberate effort in communication, relationships, and boundaries. If you do that well, remote work can unlock more opportunities than you might expect.

—Brian

Despite its critical role in maintaining a secure network, authentication software often goes unnoticed by users. Alan DeKok now runs one of the most widely used remote authentication servers in the world—but he didn’t initially set out to work in cybersecurity. DeKok studied nuclear physics before starting the side project that eventually turned into a three-decade-long career.

Read more here.

We’re just two months into 2026, and layoffs in the tech industry are already ramping up. According to data compiled by RationalFX, more than half of the 30,700 layoffs this year have come from Amazon, which announced that it would be cutting the roles of 16,000 employees in late January. Will the trend continue through 2026?

Read more here.

Recent research suggests that a majority of organizations have a significant gap when it comes to AI skills among leadership. To help fill the gap, IEEE has partnered with the Rutgers Business School to offer an online “mini-MBA” program, combining business strategy and deep AI literacy. The program spans 12 weeks and 10 modules that teach students how to implement AI strategies in their own organizations.

Read more here.

Riley Walz, a software engineer famous for his online stunts, is joining OpenAI to research and develop new ways for humans to interact with AI, WIRED has learned. An OpenAI spokesperson confirmed the hire.

Walz built a reputation as Silicon Valley’s jester and has created a series of viral web projects that double as social commentary. His most recent initiative, Jmail, lets users search Jeffrey Epstein’s emails as if they’re accessing his personal Gmail inbox. Another project, Find My Parking Cops, used publicly available data to reverse engineer San Francisco’s parking ticket system to show people exactly where each parking enforcement officer last wrote a ticket.

Now, Walz’s skills creating novel web experiences will be put to use in OAI Labs, a relatively new team led by research leader Joanne Jang. The team is secretive about what it’s been working on but has been tasked with “inventing and prototyping new interfaces for how people collaborate with AI,” according to Jang.

OpenAI has spent the past several years racing with Google and Anthropic to create new, compelling ways for people to use its AI models. While ChatGPT has been a hit with consumers, now reaching more than 800 million people every week, the company is eyeing new interfaces to improve these experiences. The move comes as millions of developers have started using coding agents such as Claude Code as their main interface to access AI models. With hires like Walz, OpenAI hopes to get ahead of the next big AI product.

Walz’s online stunts have landed him in hot water from time to time. The Find My Parking Cops website lasted just four hours before San Francisco city officials shut down the live data feed Walz’s tool relied on. A San Francisco Municipal Transportation Agency representative said at the time that it shut down the tool to ensure “employees are able to do their jobs safely and without disruption.”

It’s not always city officials giving him a hard time, though. After the CEO of UnitedHealthcare was shot dead in New York City, and police said the killer had fled on a CitiBike, Walz tried to analyze trip data he had previously scraped for a separate project to help with the search. Walz told The New York Times that people online called him a “bootlicker” for helping authorities and threatened his safety.