[AI Hanip News] Did Big Tech Just Wait for ChatGPT to Stumble... Then Launch a Full-Scale Offensive?

From Google Robots to Meta Persona
Speed Race in Expanding the Generative AI Ecosystem

Editor's NoteSince the emergence of ChatGPT, the world has been buzzing with artificial intelligence (AI). Technology is changing rapidly day by day, and news is pouring in nonstop. Even industry experts who are at the top of their game say they can hardly keep up. It’s overwhelming to consume everything, but you can’t just ignore it either. We deliver the core of weekly AI issues, carefully selected and easy to digest.

Google Revamps Everything from Search to Robots with Generative AI

Google's AI robot RT-2
[Photo by Google DeepMind Blog]

Did the comet-like rise of ChatGPT just wait for a pause? Big tech companies have launched a counterattack. Google’s response is almost like a full-scale offensive. Starting with AI search, Google is increasing AI search speed and showing not only text but also images and videos together. Speed is crucial in search, but generative AI was somewhat slow until now. AI is also being applied to YouTube. They are developing a feature where AI watches videos and automatically summarizes them. With AI summaries, it will be easier to find the perfect content even among algorithm-recommended videos. Next is the AI assistant. The assistant, which was once just a trend, is being revamped by integrating generative AI. But that’s not all. There are robots too. The new AI robot is named 'RT-2,' an upgraded version of the previous model 'RT-1.' RT-1 required new programming for each action. For example, you had to separately teach it to throw away a banana peel and a snack wrapper. In contrast, RT-2 thinks and moves on its own. Based on a large language model (LLM), once taught the concept of trash, it can distinguish and dispose of banana peels and snack wrappers as trash by itself. Google has no immediate plans to sell this robot but said it could be used as a helper for household chores or warehouse organization.

Meta Launches AI Persona... Responds in Alter Ego Style on Social Media

Meta is reportedly set to launch an AI chatbot called 'Persona' as early as next month. Given the hint that it aims to encourage users’ participation on social networking services (SNS), it seems planned for Facebook, Instagram, and Threads. As the name 'Persona' suggests, it creates a kind of 'alter ego.' Just like showing a different persona on SNS, the AI will vary its tone depending on the persona?such as an AI passionate about mukbang or an AI that keeps a proper distance from coworkers. Meta reportedly has ambitions to develop these personas into avatar chatbots that move within the metaverse.

Outline of G7 AI Regulation Proposal... First Steps for Domestic Personal Data Measures

Goh Hak-su, Chairman of the Personal Information Protection Commission
[Photo by Personal Information Protection Commission]

The outline of an international regulatory proposal for generative AI has emerged for the first time. The Japanese government has drafted the 'Hiroshima AI Process,' a G7 AI regulation proposal. The core idea is to assign a kind of certification mark from a third-party organization to articles or companies. By clicking the mark, the content creator is revealed, and even if the content is shared, the mark follows, allowing users to assess the reliability of the information. Domestically, AI guidelines have also taken their first steps. The Personal Information Protection Commission plans to establish an 'AI Privacy Team' in October to address AI-related personal data leakage issues. They will also introduce a 'prior adequacy review system' when developing AI models or services. This system allows businesses to pre-check and find ways to improve the legality and safety of data processing if needed.

Kakao: "It Will Come Out Within the Year"... KakaoTalk Transforms Once Again

Kakao Pangyo Agit, Seongnam, Gyeonggi. Photo by Jinhyung Kang aymsdream@

Kakao has finally made a firm commitment. They will not delay the unveiling of their next-generation super-large AI and plan to release it within this year (after October). To this end, they are testing AI models of various sizes that balance performance improvement and cost reduction. Since they plan to apply it to various services, they are looking for optimized models suitable for each service. The starting point will be the communication area for corporate clients. Instead of companies sending one-way messages to users on Talk Channel, AI will enable 1:1 customized conversations. Through such conversations, services like ordering, reservations, consultations, and payments are expected to become possible.

This content was produced with the assistance of AI translation services.