Mega AI such as ChatGPT requires larger datasets and complex models. Experts argue that companies can enhance their competitiveness by developing customized AI services through the acquisition of high-performance computing infrastructure capable of improving scalability by combining mega AI with cloud services.

Lee Geon-bok, Senior Vice President at Microsoft Korea, stated, “Since using ChatGPT alone is limiting, combining it with mega-AI models and specialized services is crucial,” adding, “A company’s competitiveness lies not in simple generation, but in cultivating the ability to creatively provide to customers based on cloud services.”

At a seminar held by the Korea Institute of Future Technology Education (KECFT) on the 21st at the Federation of Korean Industries building in Yeouido under the theme of 'Key Technology Issues for Mega-AI and Advanced Strategies for Commercialization,' Lee Geon-bok, Managing Director at Microsoft Korea, explained trends since the launch of mega-AI and Microsoft's Azure Open AI services.

ChatGPT is a type of generative AI and is a new AI model based on existing machine learning and deep learning algorithms. Since generative AI was developed for the purpose of communicating with humans, its learning process is essentially complete; the emergence of such AI models has spurred the need to redefine existing digital productivity.

Unlike conventional methods that processed fixed data, models, and tasks, Generative AI utilizes data sources that combine text and images as a transformer model, in addition to content corresponding to a vision within a single domain. The multimodal capability, which signifies applicability across diverse fields, holds significant meaning.

Executive Director Lee emphasized, “Although OpenAI has kept the number of parameters for ChatGPT-4 private, the exact number of parameters has become meaningless, and ultimately, it is important to consider what services can be developed through ChatGPT.”

■ MS Cloud Revenue Rises, Driven by ChatGPT Integration

Microsoft announced on the 25th that its revenue for the first quarter increased 7% year-over-year to $52.086 billion (approximately 69.6 trillion won), and its earnings per share were $2.45. Net income also rose 9% year-over-year to $18.3 billion (approximately 24.5 trillion won). This exceeded the market forecast of $51 billion.

Microsoft announced that revenue from Azure and other cloud services increased by 27%, despite forecasts that its cloud business would be sluggish due to intensifying industry competition and the economic downturn. CEO Satya Nadella stated that daily users of the search engine Bing have reached 100 million since the integration of ChatGPT, and the number of customers using Azure Open AI services has exceeded 2,500. They expressed their determination to continue investing in AI in the future.

MS began its first investment in OpenAI, the developer of ChatGPT, in 2019, and subsequently entered into an exclusive licensing agreement for OpenAI models in 2020. The key details of the exclusive licensing agreement with OpenAI for four representative services—GPT-3, Codex, DALL·E, and ChatGPT—are as follows.

Microsoft continues to make massive investments of billions of dollars in the cloud and AI sectors this year as well.

■ MS 'Azure AI Service' Equipped with ChatGPT
Microsoft provides optimized infrastructure by integrating massive AI into Azure. It officially launched the Azure Open AI service last January and added the ChatGPT preview to the Azure Open AI service last March.

Developers can leverage ChatGPT on the Azure cloud to improve existing bots, summarize call center conversations, generate new ad copy with personalized offers, and automate claims processing. The price is $0.002 per 1,000 tokens.

Microsoft's Azure AI service maximizes security compared to Open AI, providing enterprise-grade security keys and authentication, and allowing for direct management. In addition, it complies with SOC2, ISO, HIPAA, and CSA STAR compliance regulations and provides a service level of 99.95%.

In addition, from an ethical standpoint, data is collected, patterned, and continuously updated in the service based on six principles including fairness, stability, and privacy. The company explains that it has built Responsible AI, a system that blocks inappropriate questions or redefines answers through filtering based on the collected data.

Microsoft presents four steps for enterprises to develop and test services and solutions. First, collect and manage data through Azure AI services, and review related services to develop AI-based intelligent apps. Microsoft's GitHub Copilot supports developers in building environments on Azure for business enhancement and pair programming.

Executive Director Lee emphasized cloud adoption as a necessity, stating, “For companies, transitioning to the cloud to prepare and manage datasets for prompt engineering and fine-tuning will be the most fundamental preparation.” Finally, he mentioned the importance of shared value in utilizing cloud-based document management for collaboration, including various office tasks.