[ⓒSK텔레콤]
[디지털데일리 오병훈기자] Rookies who are attracting attention in the global generative AI market will gather in one place at the ‘SK AI Summit 2024’, ambitiously prepared by SK Group Chairman Chey Tae-won. As representatives of global AI startups such as OpenAI, Lambda, and Perplexity, which are leading the generative artificial intelligence (AI) craze, appear on stage as speakers, interest in them is also growing.
SK Telecom will hold the SK AI Summit 2024 at Samsung COEX on the 4th and 5th. At this event, under the theme of ‘AI together, AI tomorrow’, C-level personnel from domestic and foreign AI ecosystem companies and academic officials attended to share ways to coexist and strengthen the ecosystem in the era of general artificial intelligence (AGI). There is time to do it.
Chairman Taewon Choi is serious about AI… Inviting a large number of leading AI startups
Chairman Choi has recently repeatedly emphasized the importance of AI. Last June, during a business trip to the United States, I met OpenAI founder Sam Altman and Microsoft Chairman Satya Nadella. In the same month, at a management strategy meeting where representatives of all group affiliates gathered, he said, “There is nothing to talk about other than AI,” and ordered each affiliate to actively increase contact points with AI.
SK Group’s business is also full of AI. While SK Hynix is growing the high-bandwidth memory (HBM) business, which is the core of AI semiconductors, SK Telecom is increasing contact by establishing a collaboration system with the global search AGI company Perplexity. SK Networks also began to reorganize its business structure centered on AI. The name of the annual SK technology conference ‘SK Tech Summit’ was changed to ‘SK AI Summit’ this year.
As this is an event that Chairman Chey Tae-won pays so much attention to that he personally comes on stage, the lineup of speakers for this day also attracted attention. In particular, many representatives of leading startups in the U.S. that are leading the global generative AI market will attend. Chairman and President Greg Brockman, who founded OpenAI with Sam Ultman, Lambda CEO Stephen Balaban, and Perplexity CEO Aravind Srinivas will also appear on stage.
Greg Brockman, President and Chairman of OpenAI[ⓒ연합뉴스]
Open AI led the generative AI craze… “I came to Korea on vacation” President and Chairman Greg Brockman
At OpenAI, which is considered a leader in the generative AI model market, President and Chairman Greg Brockman (hereinafter referred to as President) will take the stage. As he is a key figure in the generative AI company that is attracting the most attention, he will have a conversation with SBVA CEO Lee Jun-pyo on the topic of ‘AI future’ immediately after Chairman Chey Tae-won’s keynote speech.
Brockman, president and chairman, studied computer science at Harvard University and the Massachusetts Institute of Technology (MIT). While in college, he stopped studying and joined the American payment platform ‘Stripe’ as an early member and began his journey as a full-fledged developer. There, he grew into a core developer and took on executive positions at the company, including being promoted to Chief Technology Officer (CTO) at an early age.
In 2015, he focused on the growth potential of AI and placed his name on the list of early founding members of OpenAI for the position of Stripe CTO. He co-founded OpenAI with Elon Musk, Sam Ultman, and Ilya Sutskever, and as a developer, he led the development of major OpenAI foundation models such as GPT-2, GPT-3, and GPT-4. Along with OpenAI CEO Sam Ultman, he is also one of the most active and active figures in external activities.
He was not an AI expert from the beginning. His explanation is that after joining OpenAI, he began accumulating AI knowledge, including machine learning, in earnest. He recalled on his blog homepage in July 2019, “During the first three years at Open AI, I dreamed of becoming a machine learning expert, but I made no progress,” and added, “Through studying mathematics and programming, I began working on a machine learning project in earnest in 2017.” “We have started,” he said.
Stephen Balaban Lambda CEO [ⓒSK AI 서밋 홈페이지 갈무리]
‘Lambda’ provides AI development infrastructure…CEO Stephen Balaban
Generative AI is one of the fields where deep learning technology is applied. It is a kind of methodology and technology that can be derived from various AI technologies. To put it simply, deep learning is a toolbox containing various tools, and generative AI is a product made with tools taken from that toolbox. Training a deep learning model involves complex calculations based on large-scale data. This is a task that cannot be handled by a typical computer. A fast and powerful supercomputer is needed. In addition, for efficient development, cloud technology that can operate the supercomputer regardless of location and time must be supported.
The company that provides this AI development infrastructure is ‘Lambda.’ Lambda’s main business is to support a deep learning-specific computing environment for AI developers’ research. It provides deep learning workstations and servers, and provides cloud-based graphics processing unit (GPU) instances to support deep learning tasks that require large-scale computation.
CEO Stephen Balaban, who founded Lambda, majored in computer science at Stanford University. He had a keen interest in computer machine learning since his college days and founded Lambda in 2012. While observing people experiencing a lack of computing resources for AI research, we came up with a project to build infrastructure to solve this problem. While leading Lambda, he focused on developing user-friendly hardware and software solutions and lowering the threshold for AI research and development.
As the head of a company that provides a ‘cradle’ for AI, he emphasized the importance of development accessibility and cooperation. In particular, we placed great emphasis on cooperation with the open source community and believed that the AI ecosystem should be developed through sharing. The view is that technology should be provided equally and democratically to everyone, and that through this, multiple ideas can be brought together to achieve sustainable growth.
If we go back in time and look at his actions, we can understand what exactly he means by ‘open source cooperation’. He developed a facial recognition application program interface (API) that operates on ‘Google Glass’, a wearable PC launched by Google in 2013. When you wear Google Glass with the corresponding API and look at the other person, information about the other person is provided in real time based on information such as SNS.
He released the API to the open source community. However, controversy soon arose that the API violated Google’s security policy. Naturally, Google officially banned the use of the API. This is an episode that clearly demonstrates the free-spirited development philosophy of CEO Balaban.
Aravind Srinivas CEO of Perplexity [Ⓒ SK텔레콤]
Purplexity attracts attention with fine tuning in the search market… Aravind Srinivas CEO
As competition for the development of foundation large language models (LLM) has intensified, much attention is being paid to fine-tuning technology that specializes in LLM functions. Foundation LLM has wide versatility, but in order to highlight specific functions, it must go through a fine-tuning process appropriate for each field. Currently, the main utility of AI models lies in ‘search’, so various LLMs that have undergone search-specific fine-tuning are attracting attention.
Perplexity introduced a search AI service that simultaneously utilizes various foundation LLMs, including the Open AI ‘ChatGPT’ series and Enthropic ‘Claude’. Through fine-tuning, Perplexity focused on improving search quality, including reducing hallucinations that reduce reliability while being based on the latest data.
Aravind Srinivas, CEO of Purplexity, was born in India and earned a doctorate in computer science from the University of California, Berkeley. Since his days as a student, he has shown the typical ‘AI geek’ aspect by presenting papers at various international academic journals and conferences through research encompassing various fields such as deep learning, reinforcement learning, and natural language processing.
He founded Perplexity in 2022 to develop an advanced information search service based on his research. Similar to Lambda CEO Balavan’s emphasis on technology accessibility, CEO Srinivas also placed importance on information accessibility.
Increasing knowledge accessibility through AI and overcoming limitations in information search are the main goals of his perplexity business. The interpretation is that the role of search AI is to ensure that users can obtain the information they need accurately and efficiently in a modern society where information is abundant and flowing.
Meanwhile, Perplexity is one of the major companies that has established a close collaborative relationship with SK Telecom. The two companies are collaborating with the goal of joint development of an AI agent (AI secretary). Last June, SK Telecom announced its investment of $10 million (approximately KRW 13.5 billion) in Purplexity. In addition, specific cases of mutual cooperation are being expanded, such as providing ‘Purplexity Pro’ vouchers to SKT users.