Bilingual

U.S. Policies Unintentionally Accelerated China's Open AI Ecosystems
21 days ago
- U.S. policies aimed at AI leadership raised costs for China's AI development but increased the strategic value of open, adaptable AI systems.
- China integrated open-source AI into national strategy through ecosystem building, standards, and resilient deployment after U.S. export controls.
- Chinese developers increased engagement with open-source LLM repositories more than U.S. developers, shifting toward open infrastructure under constraints.
- Chinese open models spread through open-source communities and research, with U.S. commercial entities using them in open-access research.
- Technological containment policies may unintentionally accelerate open innovation ecosystems as a competitive response, affecting global AI leadership.
New secrets from ancient Vesuvius scrolls
20 days ago
- Ancient papyrus scrolls from Herculaneum, buried by the Vesuvius eruption in 79 A.D., have been virtually unwrapped using AI and particle accelerator technology, revealing over 70 columns of text.
- The Vesuvius Challenge, launched in 2023, led to the discovery of new books from ancient Rome, including an eight-book series by philosopher Philodemus, with only one previously known.
- Researchers have deciphered entire scroll arguments instead of fragments, marking a "transformational shift" in accessing texts that were physically preserved but intellectually inaccessible for nearly 2,000 years.
- Despite progress with one fully unwrapped scroll and others partially decoded, more than 600 scrolls remain unread, requiring ongoing collaboration from experts in reading, editing, and understanding the ancient language.
- The project highlights interdisciplinary work combining advanced imaging, AI, and academic research to hear "voices that have been silent for 2,000 years," with the goal of deciphering the entire ancient library.
Amazon will stop accepting new customers for Mechanical Turk on July 30
20 days ago
- Amazon Mechanical Turk will close to new customers on July 30, 2026, but existing users can continue with no new features added.
- The service, launched in 2005, paid workers small amounts for tasks resistant to automation, like CAPTCHA completion or sentiment analysis.
- It has been central to debates on crowdsourced labor ethics and was used in the Facebook-Cambridge Analytica scandal.
- From 2018, it was marketed for data annotation to train neural networks as part of Amazon's SageMaker AI service.
- Mechanical Turk enabled 'fake-it-till-you-make-it' AI approaches, where human workers powered products marketed as AI, mirroring its namesake hoax.
- A 2023 analysis found 33%-46% of workers used large language models for tasks, raising data reliability and human-in-the-loop questions.
- After Amazon's decision, some on Reddit claim the platform declined years ago due to bots and fraud, predicting eventual full shutdown.

Hasty Briefsbeta

artificial-intelligence

U.S. Policies Unintentionally Accelerated China's Open AI Ecosystems

New secrets from ancient Vesuvius scrolls

Amazon will stop accepting new customers for Mechanical Turk on July 30