Quantum leap: A computer achieves error correction at record efficiency

A new quantum computer has crossed a threshold that's been eluding the field for years. Helios contains 98 physical qubits but can deliver 48 logically error-corrected qubits—a 2:1 ratio that's unprecedented. To put that in perspective: most competitors need dozens to hundreds of physical qubits just to create one logical qubit. It's the difference between building a bridge that mostly stays up and one that actually does.

The breakthrough matters because quantum computers are only useful if they can correct their own errors. Raw qubits are fragile, error-prone things. Logical qubits are the polished version—stable enough to actually compute something meaningful. Prineha Narang, a UCLA physicist and venture-capital partner, called the efficiency "unique and impressive." In a field where progress is often measured in incremental gains, this is the kind of jump that changes the conversation.

Progress across the board

Elsewhere in computing, researchers at Cornell University have miniaturized neural implants to the point of absurdity. Their wireless, laser-powered brain implant is small enough to fit on a grain of salt, yet it transmitted data about brain activity in mice for over a year without a battery replacement. The engineering required to pack that much function into something you can barely see suggests we're entering a new era of medical devices—ones that work inside your body without the burden of regular surgery or power management.

Meanwhile, one AI language model has begun parsing human language with a sophistication that rivals a graduate student in linguistics. While most large language models fail at basic grammatical analysis, this one could diagram sentences, untangle multiple meanings, and handle recursive structures the way linguists do. It's a reminder that these systems are getting better at reasoning, not just pattern-matching.

There's also the small matter of DARPA reshuffling the quantum computing landscape. The agency's QBI validation program—the first real government assessment of which quantum approaches are actually viable—has bolstered some incumbents while quietly devastating a few industry pioneers. For a field that's absorbed billions in investment, this kind of validation matters. It's the difference between hype and direction.

Beyond the lab

The practical applications are starting to arrive. Waymo is bringing autonomous robotaxis to San Diego, Las Vegas, and Detroit, accelerating a timeline that once seemed perpetually "five years away." And researchers have engineered a fungus that mosquitoes find irresistible—it killed 90 to 100 percent of them in lab tests, offering a potential new tool for disease control without the broad environmental impact of chemical sprays.

Even our space ambitions are being recalibrated. While Mars has captured the imagination, the moon is emerging as the more practical first target for terraforming efforts. It's closer, more accessible, and offers a testing ground for the technologies we'd eventually need on Mars.

One note of caution: a new study from Oxford researchers suggests that many of the benchmarks we use to measure AI progress are unreliable. Those headlines about AI passing the bar exam or achieving PhD-level intelligence? The tools used to assess those claims are often misleading. It's a useful reminder that progress isn't always as dramatic as the headlines suggest—but the actual progress, when you look closer, is real.

As neurotechnology advances, UNESCO has adopted the first global standards for protecting neural data, anticipating a future where brain-computer interfaces become routine. The 100+ recommendations range from privacy protections to guarding against scenarios that still sound like science fiction—like companies using neurotechnology to market to people in their dreams.

And in a remarkable historical footnote: archaeologists discovered that ancient hominins used the same stone tool design for 10,000 generations—hundreds of millennia—as their world changed around them. Sometimes the oldest stories teach us the most about adaptation and resilience.