Anthropic, the company behind the Claude chatbot, has unveiled Claude Science, an AI platform designed to assist researchers in analyzing data and managing complex workflows. More significantly, Anthropic has announced its intention to launch its own preclinical drug discovery programs, focusing on neglected diseases.

This ambitious move positions Anthropic not just as a technology provider but also as a direct competitor in the pharmaceutical industry, aiming to leverage large language models (LLMs) and generative AI to drastically reduce the time required for scientific tasks. While the company has not yet specified which treatments it will pursue or how it plans to navigate clinical trials, the goal is to accelerate the discovery process.

Experts caution that while AI can significantly aid in identifying promising molecules and prioritizing hypotheses, it does not eliminate the need for real-world experiments to validate efficacy, toxicity, and safety. The path from an AI-designed molecule to an approved drug remains long and complex, requiring extensive experimental validation and regulatory approval.

👉 Read the full article on Gizmodo en Español

Amazon Leo, formerly known as Project Kuiper, has achieved the critical mass of satellites required to begin its commercial satellite internet service. With the latest Atlas V rocket mission deploying an additional 29 satellites, the constellation now comprises nearly 400 units in low Earth orbit. Amazon plans to initiate service this year, starting in regions near the poles and gradually expanding towards the equator.

The project aims to provide low-latency internet access to underserved areas, directly competing with Starlink. However, Starlink currently holds a significant advantage with approximately 10,000 satellites and extensive commercial experience, while Amazon is still scaling towards its goal of over 3,000 satellites.

Amazon Leo will offer various antenna options, including Leo Nano for speeds up to 100 Mbps and Leo Ultra for enterprise and government clients, promising up to 1 Gbps download and 400 Mbps upload speeds. A key strategic advantage for Amazon could be the deep integration of Leo with AWS, offering not just internet connectivity but also direct access to cloud services for businesses and remote operations.

👉 Read the full article on Gizmodo en Español

A team of scientists from the University of California, Berkeley, and NTT Communication Science Laboratories in Japan has developed an AI system capable of translating brain activity into text. This groundbreaking technology combines functional magnetic resonance imaging (fMRI) with two AI models to describe with astonishing accuracy what a person sees, remembers, or imagines.

The system works by first analyzing thousands of video subtitles to extract semantic patterns, and then comparing these patterns with participants' brain signals to associate thoughts with linguistic structures. This AI has successfully generated text descriptions, such as "a person jumps over a waterfall in a mountain," solely from an individual's brain activity.

Neuroscientist Alex Huth noted that the model predicts with surprising detail not only visual thoughts but also captures abstract mental representations, including imagination. While this technology holds immense medical potential for individuals with conditions like aphasia, paralysis, or neurological injuries, it also raises significant ethical concerns regarding the privacy of the mind.

👉 Read the full article on Gizmodo en Español