[People人] 'Born in 1988' Kakao Brain's Kim Il-du, Drawing a New Future for Healthcare and Education with AI

Team Leader to CEO in April This Year: Bold Appointment
Leadership Concerns Despite Visible Results
Following the Release of Mega AI, AI Applied to Healthcare and Education

[Asia Economy Reporter Kang Nahum] ‘A Kakao affiliate CEO born in 1988.’

Last April, when news broke that Kim Ildu, head of the deep learning algorithm research team at Kakao Brain, would be appointed as the company’s CEO, the IT industry was abuzz. There was a mix of surprise at the bold decision by Kakao Chairman Kim Beom-su to place a team leader in his 30s at the helm of a subsidiary and concerns about the new CEO’s leadership.

Founded in February 2017, Kakao Brain is a Kakao subsidiary with about 60 employees specializing in fundamental artificial intelligence (AI) technology research. Although small in scale, the emergence of a CEO in his 30s is truly groundbreaking given the company’s focus on AI, a core industry of the future.

However, looking at CEO Kim’s background, the term ‘bold appointment’ seems understated. He joined Kakao in 2012 as a software engineer and has about seven years of experience researching and developing various AI-related services. Since 2018, he has been part of the deep learning algorithm research team as an AI engineer, demonstrating his development capabilities by working on diverse AI technologies such as computer vision, data augmentation, and medical diagnosis. Over the past three years, he has published more than ten papers in prestigious international conferences and won eight awards in international AI and machine learning competitions.

His notable research and awards include work in neural information processing systems and medical diagnosis. Last year, his latest data augmentation technology research presented at the Neural Information Processing Systems conference ‘NeurIPS’ was highly praised by many AI researchers for reducing training computation time by more than 60 times. In the same year, he won the ‘LNDb (Lung Nodule Database) Grand Challenge,’ a competition for early lung cancer diagnosis using chest CT (computed tomography) image classification algorithms, proving that AI technology can be applied to solve problems in other industries.

Upon his appointment, CEO Kim defined this year as ‘Kakao Brain Season 2,’ marking a new leap forward for the company. He expressed his ambition to accelerate AI research and aggressively lead new business growth by fully leveraging his engineering and research capabilities.

Eight months into his tenure, tangible results are emerging. Kakao Brain recently unveiled the massive AI models ‘KoGPT’ and ‘minDALL-E.’ KoGPT is a large-scale Korean language model. MinDALL-E is an image generation model that creates desired images in real-time based on user text commands. It was pretrained on a dataset of 14 million text-image pairs and has 1.3 billion parameters.

In particular, Kakao Brain is focusing on enhancing the performance of the KoGPT model. By utilizing Google’s Tensor Processing Unit (TPU), a specialized chip for tensor operations, they introduced the largest deep learning supercomputing infrastructure in Korea, surpassing 1 exaFLOPS (a computing speed unit representing 10^18 operations per second), improving research efficiency. Additionally, they increased the KoGPT model’s parameters from 6 billion to 30 billion, making it five times larger and enabling it to execute commands more accurately than before.

Currently, CEO Kim’s attention is focused on ‘healthcare’ and ‘education.’ At a press conference held on the 20th, he said, "We aim to challenge ourselves so that AI becomes a technology that changes the world," adding, "Among these, healthcare and education are high value-added sectors."

In this regard, Kakao Brain is currently developing AI technology to predict protein structures. They also made a 5 billion KRW investment in the AI drug design platform company ‘Galaxy.’ Drug development typically takes an average of five years just to develop candidate substances and costs about 670 million USD (approximately 790 billion KRW), with a success rate below 10%, making it a high-risk field. Kakao Brain plans to significantly reduce drug development time and costs through AI. They are also considering developing AI to assist language education and coding education.

CEO Kim aims to expand fundamental AI technology research next year and develop AI services with significant social impact. He stated, "Kakao Brain will lead differentiated research technologies and ensure these technologies are applied to important real-world problems," adding, "We are researching to realize human-like AI based on about 2 billion image-text data pairs, and we are also considering ways to partially open this to contribute to the overall AI industry."