Frontier language models lack dnderstanding of the world, new technique for all-optical computing, and new affortable cancer detection solution [11-06-2024]

University of Rochester Scientists Develop Fast, Affordable Cancer Test from a Single Drop of Blood

Researchers at the University of Rochester have developed a rapid and cost-effective cancer test using ultrathin membranes to capture extracellular vesicles from a single drop of blood. It can also identify critical immune modulatory proteins, offering potential for predicting patient-specific selection of immunotherapies.

Source: https://scitechdaily.com/scientists-develop-fast-affordable-cancer-test-from-a-single-drop-of-blood/

Researchers Find that Despite Impressive Output, Generative AI Models Lack Coherent Understanding of the World

In a recent study, researchers found that despite their impressive output, generative AI models do not have a coherent understanding of the world's rules, which can lead to unexpected failures in similar tasks. This has significant implications for deploying these models in real-world scenarios, as the models may perform well in one context but break down if the task or environment slightly changes.

Source: https://news.mit.edu/2024/generative-ai-lacks-coherent-world-understanding-1105

'Diffraction Casting' Method for All-Optical Logic Operations Computing

Innovators at Tokyo University have developed a new 'diffraction casting' method which uses light to achieve faster and more efficient data processing. This method marks a significant step towards revolutionizing the computing industry, offering scalability, parallel logic operations, and high flexibility and integration capability. The research highlights the potential of optical computing using spatial parallelism as a building block for future computing systems, paving the way for a new information processing framework.

Source: https://scitechdaily.com/parallel-and-multiplexed-the-new-wave-of-all-optical-logic-operations/