[AI One-Bite News] Nuclear Development and Self-Replication... AI Disaster Response Team Emerges

OpenAI Establishes Team to Analyze and Respond to AI-Related Disasters

Editor's NoteSince the emergence of ChatGPT, the world has been buzzing with artificial intelligence (AI). Technology is changing rapidly, and news is pouring in every day. Even top industry experts are said to be struggling to keep up. It's overwhelming to consume it all, but you can't afford to miss out either. We deliver the core of weekly AI issues, carefully selected and easy to digest.

Apple, Late to the Game, Accelerates Generative AI

The prosecution has launched an investigation into Apple on suspicion of overcharging thousands of billions of won in in-app payment fees from domestic application developers. The photo shows the Apple Store Gangnam in Seoul on the 18th. Photo by Jinhyung Kang aymsdream@

Apple, notably slow in the generative AI field, is making a comeback. It is reported that as early as next year, Apple plans to integrate generative AI like ChatGPT into its own products. Reports also suggest an annual investment of $1 billion (approximately 1.35 trillion KRW) to catch up with competitors. While Apple uses AI for photo and text editing in some products, it has not launched generative AI services. It is only known to have internally developed a chatbot service similar to ChatGPT using its own large language model (LLM). Next year, Apple may integrate generative AI into its AI voice assistant Siri, iPhone operating system (OS), and internal development tools.

"Preventing AI Disasters"… OpenAI Forms Risk Response Team

OpenAI has created a new team to respond to 'catastrophic risks' related to AI. This team analyzes and prevents risks including chemical and biological threats like nuclear development, as well as AI's self-replication behaviors. It also covers AI that deceives humans and cybersecurity threats. OpenAI CEO Sam Altman, along with prominent AI researchers, issued a statement last May emphasizing that "the world must prioritize reducing the risk of human extinction caused by AI." In an interview, he also warned that governments should treat AI as seriously as nuclear weapons.

China's AI Patent Applications Double Those of the U.S.

[Image source=Yonhap News]

China significantly outpaced the U.S. in AI patent applications last year. China filed 29,853 AI-related patents, ranking first worldwide. The U.S., in second place, filed 16,805 patents, about half of China's number. Japan (8,870) and South Korea (7,899) followed. Since taking the lead in 2017, China has been widening the gap with the U.S. While U.S. AI patent applications have been declining annually, China, backed by government support, continues to increase its patents.

"Developing AI That Thinks Like Humans"

An AI that thinks like a human has been developed. This AI was created by a team led by Brandon Lake, a psychology professor at New York University. They applied a 'meta-learning' technology that mimics human thinking. Through this technology, once the AI learns one concept, it can combine it with other concepts to expand its knowledge. For example, after learning the word 'running,' it understands phrases like 'running twice around the room' or 'running with hands raised.' Previously, it was widely believed that AI could not perform such thinking, and attempts to develop it failed. In contrast, this newly developed AI was intensively trained to create new word combinations from a single word. While it lags behind ChatGPT in general learning tasks, it outperformed ChatGPT in meta-learning tasks.