Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 50

Hello Universe: NASA’s Next-Gen Space Processor Undergoes Testing

NASA’s High Performance Spaceflight Computing project aims to dramatically improve the computing power of spacecraft. Missions need processors that can withstand the harsh space environment, so they use chips developed years ago that are hardy and reliable. But upgraded chips are needed to enable the development of autonomous spacecraft, accelerate the rate of scientific discovery through faster data analysis, and support astronauts on missions to the Moon and Mars.

“Building on the legacy of previous space processors, this new multicore system is fault-tolerant, flexible, and extremely high-performing,” said Eugene Schwanbeck, program element manager in NASA’s Game Changing Development program at the agency’s Langley Research Center, in Hampton, Virginia. “NASA’s commitment to advancing spaceflight computing is a triumph of technical achievement and collaboration.”

The centerpiece of the High Performance Spaceflight Computing project is a new radiation-hardened, high-performance processor, designed to provide up to 100 times the computational capacity of current spaceflight computers while enduring a barrage of challenges in space. NASA’s Jet Propulsion Laboratory in Southern California has been conducting various tests that replicate those challenges.

It took 40 years for technology to catch up to this zipper design

In 1985, the Innovative Design Fund placed an ad in Scientific American offering up to $10,000 to support clever prototypes for clothing, home decor, and textiles. William Freeman PhD ’92, then an electrical engineer at Polaroid and now an MIT professor, saw it and submitted a novel idea: a three-sided zipper. Instead of fastening pants, it’d be like a switch that seamlessly flips chairs, tents, and purses between soft and rigid states, making them easier to pack and put together.

Freeman’s blueprint was much like a regular zipper, except triangular. On each side, he nailed a belt to connect narrow wooden ‘teeth’ together. A slider wrapping around the device could be moved up to fasten the three strips into place, straightening them into a triangular tube. His proposal was rejected, but Freeman patented his prototype and stored it in his garage in the hopes it might come in handy one day.

Nearly 40 years later, MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers wanted to revive the project to create items with ‘tunable stiffness.’ Prior attempts to adjust that weren’t easily reversible or required manual assembly, so CSAIL built an automated design tool and adaptable fastener called the ‘Y-zipper.’ The scientists’ software program helps users customize three-sided zippers, which it then builds on its own in a 3D printer using plastics. These devices can be attached or embedded into camping equipment, medical gear, robots, and art installations for more convenient assembly.

A new system developed at MIT CSAIL helps users design three-sided fasteners called “Y-zippers,” then 3D prints them. The devices can be attached or embedded to camping equipment, medical gear, robots, and art installations, seamlessly switching each item between soft and rigid.

A global screen for magnetically induced neuronal activity in the pigeon brain

What if every scientific paper you read was just the “highlight reel” of a much longer, messier, and more complicated movie? You see the breakthrough, but you never see the hundreds of hours of footage showing what didn’t work.

Ultimately, the ARA marks a shift toward a future where “The Last Human-Written Paper” isn’t the end of science, but the beginning of a much deeper, machine-readable conversation.

However, this shift toward radical transparency comes with its own set of hurdles. While ARAs make AI agents more efficient, the study found a “prior-run box” effect where seeing a human’s past failures actually limited an AI’s ability to think outside the box and find creative new solutions. There is also a significant cultural and technical gap to bridge: the system relies on researchers being willing to expose their “messy” unfinished work, and even with better data, the jump in actual experiment reproduction was relatively modest. Furthermore, the reliance on “compilers” to translate old papers into this new format risks baking in errors or “hallucinations” if the original source was vague, proving that while machine-readable data is powerful, it isn’t a magic fix for the inherent complexities of scientific discovery.

How animals detect Earth’s magnetic field remains a mystery in sensory biology. Despite extensive behavioral evidence, the neural circuitry and molecular mechanisms responsible for magnetic sensing remain elusive. Adopting an unbiased approach, we used whole-brain activity mapping, tissue clearing, and light sheet microscopy to identify neuronal populations activated by magnetic stimuli in the pigeon (Columba livia). We demonstrate robust, light-independent bilateral neuronal activation in the medial vestibular nuclei and the caudal mesopallium. Single-cell RNA sequencing of the semicircular canal cristae revealed specialized type II hair cells that express the molecular machinery necessary for the detection of magnetic stimuli by electromagnetic induction.

Performance of a large language model on the reasoning tasks of a physician

What if every scientific paper you read was just the “highlight reel” of a much longer, messier, and more complicated movie? You see the breakthrough, but you never see the hundreds of hours of footage showing what didn’t work.

Ultimately, the ARA marks a shift toward a future where “The Last Human-Written Paper” isn’t the end of science, but the beginning of a much deeper, machine-readable conversation.

However, this shift toward radical transparency comes with its own set of hurdles. While ARAs make AI agents more efficient, the study found a “prior-run box” effect where seeing a human’s past failures actually limited an AI’s ability to think outside the box and find creative new solutions. There is also a significant cultural and technical gap to bridge: the system relies on researchers being willing to expose their “messy” unfinished work, and even with better data, the jump in actual experiment reproduction was relatively modest. Furthermore, the reliance on “compilers” to translate old papers into this new format risks baking in errors or “hallucinations” if the original source was vague, proving that while machine-readable data is powerful, it isn’t a magic fix for the inherent complexities of scientific discovery.

We systematically evaluated the medical reasoning abilities of an LLM across six diverse experiments, comparing the model with hundreds of expert physicians. Overall, the model outperformed physicians across experiments, including in cases utilizing real and unstructured clinical data taken directly from the health record in an emergency department. These diagnostic touchpoints mirror the high-stakes decisions taken in emergency medicine departments, where nurses and clinicians make time-sensitive choices with limited information. Our results showed that humans, GPT-4o, and o1 all improved their diagnostic abilities as more information was available; o1 outperformed humans at multiple touchpoints, with the widest gap at initial ER triage, where there is the least information available.

The rapid pace of improvement in LLMs has substantial implications for the science and practice of clinical medicine. Although applying AI to assist with clinical decision support is sometimes viewed as a high-risk endeavor (22, 23), greater use of these tools might serve to mitigate the human and financial costs of diagnostic error, delay, and lack of access (24, 25). Our findings suggest the urgent need for prospective trials to evaluate these technologies in real-world patient care settings and for health care systems to prepare for investments for computing infrastructure and design for clinician-AI interaction that can facilitate the safe integration of AI tools into patient-care workflows. This includes the development of robust monitoring frameworks to oversee the broader implementation of AI clinical decision support systems (22), monitoring not just final diagnostic accuracy but other metrics crucial for successful deployment, including safety, efficiency, and cost.

We emphasize that our study addresses only text-based performance for both humans and machines; clinical medicine is multifaceted and awash with nontext inputs, including auditory (such as the patient’s level of distress) and visual information (for example, interpretation of medical imaging studies) that clinicians routinely use. Existing studies suggest that current foundation models are more limited in reasoning over nontext inputs (26, 27); future work is needed to assess how humans and machines may effectively collaborate (28) in use of nontext signals. This requires new benchmarks, trials, and technological solutions to more faithfully measure clinical encounters. Existing investment in increasingly pervasive ambient AI scribes and other passive monitoring technologies holds promise to serve as the basis for such investigations.

Continue reading “Performance of a large language model on the reasoning tasks of a physician” | >

Hiring a Futurist? The Red Flag Most Leadership Teams Miss

When a leadership team hires a futurist to think about the future for them, the hire itself is the failure.

I say this as someone who gets booked to think about the future.

Executives increasingly hire futurists, consultants, and now AI tools to handle their thinking about what comes next. It feels like rigor. It looks like preparation. It is, in fact, an abdication.

Nobody can tell you what is coming with the precision your strategy deck assumes. Not a futurist. Not a consultant. Not an AI system trained on every word ever written. Plenty of people sell certainty about the future. Nobody can actually deliver it.

So how do you tell whether you are using a futurist well, or whether the booking itself is the warning sign?

I wrote a short piece arguing that the answer comes down to one question every executive should ask before signing the contract.

Full piece: [ https://www.singularityweblog.com/hiring-a-futurist/](https://www.singularityweblog.com/hiring-a-futurist/)

Continue reading “Hiring a Futurist? The Red Flag Most Leadership Teams Miss” | >

Metastatic cancer detection and management with artificial intelligence and augmented reality (Review)

Metastatic cancer remains a significant global health challenge, contributing to the majority of cancer-related mortality due to late detection, therapeutic resistance and the complexity of disseminated disease. Recent advances in artificial intelligence (AI) and augmented reality (AR) are transforming the landscape of metastatic cancer detection and management. AI-driven tools, including radiomics, deep learning models, and predictive analytics, enhance early identification of metastatic lesions, improve diagnostic accuracy, and support personalized treatment strategies by integrating multimodal clinical, imaging and molecular data. At the same time, AR technologies are increasingly applied in image-guided surgery, real-time tumor visualization and patient education, enabling more precise interventions and improved clinical decision-making.

Ultrafast switching device unlocks low-power optical-to-electrical conversion for AI hardware

Modern energy demands are soaring as technologies like AI and IoT become more common, and researchers have been working hard to develop hardware that can keep up. Now, a team of researchers from the University of Tokyo has developed an ultrafast and energy-efficient nonvolatile switching device, described in an article published in the journal Science, that may soon be able to significantly reduce power consumption for high-energy demand technologies.

Currently, most nonvolatile switching devices for data processing architectures have operating speeds in the nanosecond range. However, faster speeds are required for modern central processing units (CPUs), which operate in the gigahertz range.

At 5 GHz, a single cycle lasts only 200 picoseconds. If a switching device takes a nanosecond (1,000 picoseconds) to turn on or off, it misses multiple clock cycles, creating a major bottleneck that prevents the processor from operating continuously at full capacity. Optical interconnects are being explored to overcome electronic bottlenecks, but more efficient optical-to-electrical (O/E) conversion is still needed.

/* */