Summary
Highlights
The U.S. Treasury Department, President Joe Biden, and former President Donald Trump are concerned about China's abundant and low-cost electric vehicles and solar panels. Treasury Secretary Janet Yellen notes these products are exported at artificially low prices due to government support and cheap labor, causing U.S. companies like Soneva to struggle. Economist Eswar Prasad highlights China's need to export these products due to domestic overcapacity. The U.S. previously imposed tariffs on Chinese steel and aluminum, and similar actions might be considered for green energy products. China argues its low-cost green products benefit the world and points to U.S. incentives for electric vehicles as a violation of WTO rules. However, a study by the Center for Strategic and International Studies found China's financial aid to its green industry in 2019 was more than double that of the U.S. Yellen warns against flooding the U.S. market with Chinese products and emphasizes the U.S.'s concern over China's industrial strategy.
OpenAI has introduced 'Voice Engine,' a tool capable of reproducing human voices from a 15-second audio sample and written instructions. Developed since late 2022, its public release is delayed due to safety concerns. Testing with trusted partners aims to refine safety and explore beneficial applications. Publicized instances of fake voice recordings, including one impersonating President Biden to dissuade voters, highlight misuse risks, especially during election years. Deepfakes, which create audio or video to show people saying or doing things they didn't, are a growing concern. OpenAI is discussing ways to mitigate these risks and combat financial crimes, such as using cloned voices for fraud. The company is collaborating with various sectors to guide development and deployment. Testers must obtain permission to represent voices and disclose AI generation. Other companies like Descript also offer voice cloning services, with Descript claiming 60-second audio samples suffice. OpenAI envisions positive applications, including reading assistance, support for non-speaking individuals, and voice recovery, with multilingual capabilities allowing voice and style preservation across languages.
Brian Lin discusses OpenAI's Voice Engine with Ashley Thompson, elaborating on its ability to reproduce human voices with high quality. OpenAI has released samples but is delaying public release due to safety concerns, particularly regarding the potential for generating fake audio of politicians during elections and unauthorized voice cloning. Lin notes that other voice cloning tools exist, mainly targeting audio production and podcasters, while OpenAI also focuses on educational and speech-support applications like assisting non-readers or those with speech difficulties, and even translating spoken English into other languages while maintaining the original speaker's voice and style.
Ana Mateo, Andrew Smith, and Jill Robbins introduce a new segment for "Let's Learn English" featuring Ana's work and life in Washington, D.C. In lesson 22, Ana and her producer Amelia plan a children's show, discussing fun summer activities. They imagine segments on camping, hiking, fishing, amusement parks, and beaches. Andrew shares his love for hiking in the mountains and clarifies the difference between 'hiking' and British English 'trekking.' Jill enjoys swimming, visiting family, and painting. The lesson also explores grammar, specifically three ways to express enjoyment: "like to + base verb," "like + -ing form (gerund)," and "like to go + gerund," providing numerous examples of activities that use "go + gerund." They demonstrate the quick pronunciation of "like to" as "like tuh" and present a chant to practice common activities, emphasizing that having fun with language aids learning. The segment concludes by differentiating between 'fun' (enjoyable activity) and 'funny' (something that makes you laugh).