Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI

New AI tool learns to read medical images with far less data

A new artificial intelligence (AI) tool could make it much easier—and cheaper—for doctors and researchers to train medical imaging software, even when only a small number of patient scans are available.

The AI tool improves upon a process called medical image , where every pixel in an image is labeled based on what it represents—cancerous or normal tissue, for example. This process is often performed by a highly trained expert, and has shown promise in automating this labor-intensive task.

The big challenge is that deep learning-based methods are data hungry—they require a large amount of pixel-by-pixel annotated images to learn, explained Li Zhang, a Ph.D. student in the Department of Electrical and Computer Engineering at the University of California San Diego. Creating such datasets demands expert labor, time and cost. And for many medical conditions and , that level of data simply doesn’t exist.

Big Tech’s Big Bet on AI Driving $344 Billion in Spend This Year

If there’s any lesson to take from the spending plans issued by the world’s largest technology companies over the past two weeks, it’s to never underestimate the fear of missing out.

Microsoft Corp., which set a $24.2 billion capital spending record last quarter, plans to drop upwards of $30 billion in the current period. Amazon.com Inc. similarly spent $31.4 billion last quarter, almost double what it dropped a year ago, and is maintaining that level of investment. Google owner Alphabet Inc. raised its capital expenditures guidance this year to $85 billion.

Researchers harness AI-powered protein design to enhance T-cell based immunotherapies

A paper published in Cell highlights how researchers have leveraged AI-based computational protein design to create a novel synthetic ligand that activates the Notch signaling pathway, a key driver in T-cell development and function.

These so-called soluble Notch agonists can be broadly applied to optimize clinical T-cell production and advance immunotherapy development.

Notch signaling is central to many cellular differentiation processes and is essential in transforming human immune cells into T-cells that target viruses and tumors. But activating Notch signaling in the laboratory has posed a challenge.

Windows 11’s Copilot app confirms GPT-5, Microsoft 365 Copilot, Azure prepares for GPT-5

GPT-5 could begin rolling out in the next few days, if everything goes to plan. There are enough evidences to confirm that Microsoft is preparing Copilot (its consumer-facing AI assistant), Microsoft 365 (primarily tailored for businesses and work), and Azure (enterprises/API customers) for GPT-5.

GPT-5, also referred to as GPT-5 alpha in early leaked benchmarks, is OpenAI’s next SOTA (State of the Art) model, and it has the potential to disrupt the AI industry again.

One source describes GPT-5 as phenomenal in coding, and it doesn’t look like it will be rolled out to just paid consumers, as even those without a subscription will be able to access it.

Amazon backs Skild AI and its revolutionary artificial intelligence model for robots capable of learning multiple tasks

Skild AI just took the wraps off Skild Brain, a “general-purpose” model meant to run on many robot bodies —not just one factory arm or one warehouse cart. The demos weren’t sci-fi eye candy; they were the unglamorous moves that make or break real deployments: climbing stairs, recovering balance after a shove, picking from clutter. The bet is simple and bold: if you train a single model across lots of tasks and lots of robots, then every new job makes the whole system better.

To understand Skild AI’s approach, think of three streams of experience flowing into one brain.

First, millions of simulated episodes where robots practice safely at super-speed. Second, human-action videos that teach the model what skilled manipulation looks like. Third, real-world logs from customer robots running Skild AI software—those streams are fed back to refine the model so the next update is smarter on day one.