[Digital Today reporter Chi-gyu Hwang (황치규)] Publisher Insight has released a Korean translation of "The Age of Scaling," a book sharing the reality of AI innovation based on interviews with key figures at the center of the global AI field.

The publisher said author Dwarkesh Patel (드와르케시 파텔) offers a comprehensive, contemporary portrait of AI based on in-depth interviews with AI researchers and company founders including Anthropic CEO Dario Amodei (다리오 아모데이), DeepMind co-founder Demis Hassabis (데미스 하사비스), OpenAI co-founder Ilya Sutskever (일리야 수츠케버), MIRI co-founder Eliezer Yudkowsky (엘리에저 유드카우스키) and Meta CEO Mark Zuckerberg (마크 저커버그).

It covers a wide range of topics, from technical details on how LLMs work to the possibility that AI could take over the world and the potential for explosive economic growth. It digs into issues people on the front lines consider most important, including the power of scaling, the possibility of alignment failure, the massive inputs required for AGI and the economic and social repercussions of superintelligence.

The author hosts the "Dwarkesh Podcast" and has conducted in-depth interviews by inviting self-taught experts and influential figures.

According to the book, Amodei, Anthropic's CEO, says of AI, "In fact, we still don't know. It is almost entirely close to an accidental empirical fact. It is something you can detect in the data, but we still have not obtained a satisfying explanation."

Hassabis of Google said, "We have to push scaling as hard as we can. Whether we hit a wall is something we will only know when we get there. Nobody knows. In the meantime we should focus even more on innovation. Of the efforts we have made, only half are related to scaling. The other half should be investment in the architectures and algorithms we will need next time, because increasingly larger scaled models will appear. If it were me, I would bet that we need both."

Sutskever said, "What I will do after AGI and what others will do is a very tricky question. Where will people find meaning? But it is also a question AI can help with. I imagine we can become wiser because we interact with AGI. AI will help us see the world more accurately and make our inner selves better through our interactions with AI. Imagine talking to the best meditation teacher in history."

The publisher said the book can also be read as an independent introductory guide to understanding the technology itself. It includes explanations of more than 170 terms, commentary on technical issues mentioned by participants, classic essays by other authors, and previously unpublished interviews with Open Philanthropy research analyst Ajeya Cotra and Anthropic co-founder Jared Kaplan (재러드 캐플런).