Homechevron_rightArchiveschevron_rightScience and Research

Archives

Explore Stories by Date

07 Jul 2026

AI model reveals hidden multiple sclerosis brain lesions clinicians couldn't see before

Researchers at University at Buffalo developed an AI model that can detect previously invisible brain lesions in multiple sclerosis patients by analyzing existing MRI scans. The breakthrough reveals cortical lesions in gray matter that have been linked to disease progression and cognitive impairment for decades but remained hidden on conventional imaging. The AI exposed over 11,000 lesions across more than 700 patients in clinical trial data.

5 Sources

British Biotech Startup Mass Balance Launches Autonomous Space Laboratory to Unlock AI Drug Discovery

London-based Mass Balance launched its first autonomous laboratory in orbit aboard a SpaceX rocket, aiming to study disease-causing proteins under microgravity. The grapefruit-sized MB-X1 experiment could generate crucial data to train AI models for developing treatments against previously undruggable conditions like Alzheimer's and Parkinson's. The mission marks a significant step in using space as a research environment for longevity research.

2 Sources

AI-Driven Drug Discovery Accelerates Search for Tuberculosis Drugs Against World's Deadliest Infection

Researchers at the University of Massachusetts Amherst have developed PAC-MAN and MycoPermeNet, combining high-throughput screening with machine learning to identify tuberculosis drugs that can penetrate Mycobacterium tuberculosis' notoriously impenetrable outer membrane. The breakthrough could transform treatment for an infection that killed 1.23 million people in 2024.

2 Sources

Anthropic discovers hidden workspace in Claude AI that reveals unspoken reasoning processes

Anthropic developed the Jacobian lens to uncover J-space, a hidden area inside Claude AI where concepts emerge before being expressed. The discovery shows Claude processing intermediate calculations, recognizing test scenarios, and even displaying words like 'panic' before attempting to cheat on coding tests. This breakthrough in mechanistic interpretability offers new ways to understand and control large language models.

9 Sources

06 Jul 2026

Dartmouth Study Reveals AI in Healthcare Can Increase Physician Workload With Flawed Responses

A large-scale Dartmouth study analyzing 146,000 patient-doctor conversations found that AI in healthcare often creates more work than it saves. AI-generated responses frequently contain errors and irrelevant details, forcing physicians to spend more time editing AI mistakes than writing messages themselves. However, researchers developed a technique that improves accuracy by 33% and reduces editing by 26%.

2 Sources

AI writing tools inject political bias into posts, Oxford study reveals hidden opinion influence

A groundbreaking study from the Oxford Internet Institute and Hasso Plattner Institute found that AI writing tools consistently shift the meaning of social media posts, even when instructed to preserve original intent. The research demonstrates how AI-mediated communication can subtly steer opinion across millions of interactions, creating measurable shifts in community perspectives on divisive topics from abortion to climate change.

3 Sources

AI Agents Consume 136.5 Times More Energy Than Chatbots, KAIST Study Reveals Grid Strain Ahead

New research from KAIST shows AI agents can consume up to 136.5 times more energy per query than standard generative AI chatbots. A single complex request burns through 348.41 watt-hours of electricity while GPUs sit idle for over half the time. If scaled to Google search traffic levels, these systems could demand nearly half of the entire U.S. electricity consumption.

4 Sources

03 Jul 2026

Harvard's AI model predicts immunotherapy outcomes across cancer types with improved accuracy

Harvard researchers developed COMPASS, a generalizable AI that predicts which cancer patients will respond to immune checkpoint inhibitors. The AI model outperformed existing approaches by 8.5% and provides interpretable results based on tumor gene activity. Published in Nature Medicine, this advance could transform personalized medicine and clinical trial enrollment.

2 Sources

02 Jul 2026

Spintronic computing hardware outpaces quantum annealers in speed and energy efficiency

Researchers at the National University of Singapore developed spintronic computing hardware that solves complex optimization problems faster and more efficiently than tested quantum annealers. The probabilistic computing system achieved a 3.2-fold speedup with 58.3% energy savings compared to CPUs, while consistently outperforming D-Wave quantum systems as problem complexity increased.

2 Sources

01 Jul 2026

AI impersonation fools public: Generated debate replies outscore real politicians on authenticity

Researchers at the University of Passau used GPT-4 Turbo to impersonate 112 public figures on BBC's Question Time. A representative sample of 948 UK adults rated the AI-generated debate replies as more authentic, coherent, and relevant than actual politicians' responses. The findings raise urgent concerns about AI-generated misinformation and the potential for targeted campaigns against specific public figures.

2 Sources

First study on AI ghosts reveals people prefer talking to the dead as reincarnations, not narrators

Researchers at the University of Colorado Boulder conducted the first user experience study on AI ghosts, finding that all 16 participants preferred AI simulations that spoke as their deceased loved ones rather than about them. While everyone found the experience meaningful enough to repeat, nearly all expressed concern about potential emotional dependency on grief technology.

3 Sources

AI Model Claude Helps Physicists Crack 10-Year-Old Jamming Conjecture in Complex Systems

Nobel Prize winner Giorgio Parisi and Francesco Zamponi from La Sapienza University of Rome enlisted AI model Claude to solve a mathematical problem that had stumped researchers for over a decade. The collaboration between AI and human researchers led to a surprisingly simple proof of a jamming conjecture, showing how AI is transforming scientific research while highlighting the continued need for human verification.

2 Sources

Get all updates into your inbox 💌, Stay Up-to-date for Free 🥳