The past year has been a year of transformation in the field of AI, characterized by breakthrough innovations, emerging regulatory frameworks, and a shift towards practical AI tools that enhance productivity.

2023 was a year filled with awe and caution, while 2024 became the era of AI experimentation. By 2025, companies will explore the practical applications of AI in organizations more pragmatically.

2024 Review

OpenAI continues to lead the way.

In 2024, OpenAI continues to maintain its leading position in the field of AI innovation. The release of GPT-4o in May marked a milestone as a multimodal model, achieving content generation and understanding across various media forms, including text, images, and video for the first time. Subsequently, o1 and its lightweight version o1-mini were launched in September, and an upgrade to o3 was made in December.

Notably, OpenAI's Sora project has completely revolutionized video generation technology. From its release in February to the launch of the enhanced Sora Turbo in December, this technology has demonstrated an unprecedented ability to generate realistic video content from text descriptions.

AI agents and assistants

In 2024, the focus of AI shifted from replacing jobs to enhancing human productivity through innovative tools and assistants. Google launched several groundbreaking projects, including the AI assistant Astra for smartphones and smart glasses, as well as the Chrome extension Mariner that allows Gemini to interact with web browsers. These tools represent a transformation from simple chat interfaces to interactive intelligences that can understand and manipulate digital environments.

Anthropic also joined this trend, enabling its AI assistant Claude to use a computer like a human—viewing the screen, moving the cursor, and interacting with interfaces. This development opens up new possibilities for AI collaboration in everyday computing tasks.

The developer Community has also welcomed exciting advancements. GitHub Copilot and Cursor AI have received unanimous praise from developers, while automated coding intelligences like Devin and GitHub Workspace are still in the refinement stage. These tools are not intended to replace developers but rather to significantly enhance development efficiency by handling routine tasks and providing improvement suggestions.

AI Regulation

In 2024, significant progress was made in AI regulation, especially with the introduction of the EU AI Act. This milestone regulation establishes a comprehensive framework for the development and deployment of AI systems for the first time, balancing innovation while ensuring safety and ethical considerations.

Looking ahead to 2025

2024 was a year of significant advancements in the field of artificial intelligence. We witnessed numerous new applications of AI in daily life and new methods for improving models, moving us closer to AGI.

Looking ahead to 2025, the focus seems to be shifting towards more specialized and efficient AI systems, enhancing reasoning abilities through technologies like compute during testing, and concentrating on AI tools that improve productivity. These trends suggest that the future AI will not only become smarter but also closer to human needs.

Companies have begun to translate proof of concept (PoC) into production applications, with an expected increase in investment proposals for AI in the coming months. However, there is still a lack of awareness or concerns about this technology and its future. To this end, we summarize the main AI trends worth watching in the future.

AI Trends

The past two years have been a phase of adaptation and experimentation for AI, but through the exploration of this phase, companies and organizations have solidified their data strategy and become more systematic in pursuing value creation through AI.

The next goal of AI strategy is the operationalization of large language models (LLM), which requires evolution in observability, governance, and security. Finding the balance between reliability, transparency, and interpretability will be key in pursuing true corporate goals and values.

The overarching challenges related to governance, security, and compliance are closely linked to data quality and AI strategy, prompting the need for AI operationalization. Although it sounds challenging, this requires a data platform and tools centered around AI, but if companies and organizations have already established a unified data strategy, then half the work is done. The other half depends on AI tools that meet specific application goals and building a corporate culture around the technology.

AI and Machine Learning: Key Trends and Developments for 2025

The coming months will mark the transformation of AI from a mere tool to an indispensable part of work and personal life.

AI-driven agents will possess greater autonomy, simplifying our lives while becoming new ways to tackle major challenges.

The adoption rate of generative AI has increased from 55% to 75% over the past year, and new AI tools will further unleash its potential.

AI models will become more powerful and practical.

We have witnessed an ongoing increase in the speed and efficiency of AI models when performing tasks such as programming or writing text from scratch. This trend will only advance further by 2025.

Models with advanced reasoning capabilities can solve complex problems through logical steps, just like humans do when thinking about intricate issues. This is particularly evident in fields such as programming, mathematics, science, medicine, and law, enabling them to generate code, compare contracts or documents, and execute multi-step workflows.

These advancements not only drive model innovation but also improve data management and post-training processes. As models become faster, better, and more specialized, they will give rise to entirely new AI experiences and higher practicality.

AI agents will revolutionize the way we work.

AI agents will be a major technological trend in 2025. For example, data from Microsoft shows that about 70% of Fortune 500 companies are already using Microsoft 365 Copilot to complete many repetitive and routine tasks, such as checking emails and taking meeting notes.

By 2025, a new generation of AI agents will surpass these tasks, benefiting from advancements in memory, reasoning, and multimodal capabilities.

Companies can reimagine processes, such as generating HR reports, resolving equipment issues, or answering customer queries, thereby freeing employees from tedious tasks and allowing them to focus on more valuable work.

In fact, companies can even configure multiple AI agents to alert supply chain managers about inventory disruptions, recommend new suppliers, and execute sales orders. Moreover, the breakthrough lies in creating and using agents without extensive technical skills, such as Microsoft Copilot.

More efficient resource management.

Although AI requires consumption of resources such as Energy, it can also serve as a tool to address this issue.

In the coming years, new types of AI Datacenters will be implemented that do not use water for cooling while expanding the application of efficient liquid cooling systems.

There will also be investments in carbon-free Energy to provide more carbon-free Electrical Utilities for the operation of these AI centers.

Building responsible AI through customization.

By "measuring," it is possible to define and assess the risks of AI, which is crucial when developing responsible AI. Therefore, a significant breakthrough in 2025 will be testing and personalization.

To achieve AI that is free from bias, privacy, or security issues, rigorous and comprehensive testing is required. If risks and threats can be measured, they can be addressed or mitigated.

Moreover, even if models become safer, testing and measurement still need to reach a level that can handle the most severe potential threats.

More importantly, personnel within enterprises and organizations will have greater control over how AI applications operate. They can customize applications to filter content and set limitations appropriate for their work.

Accelerating Scientific Progress

We have already seen the immense impact of AI on multiple Global Industries. One of these is driving the development of scientific research and unlocking new capabilities in natural sciences, sustainable materials, drug discovery, or treatment.

By 2025, the application of AI in research could drive progress in addressing some of the world's most pressing issues, such as climate change and various diseases.

Trends in Generative AI

According to Gartner's report "2024 Generative AI Hype Cycle," the following four major trends are receiving the most attention in the industry:

1. AI agents: Unlike traditional chatbots, these models are becoming more proactive and efficient. To meet our expectations, significant progress must be made in reasoning, memory, and the ability to remember and contextualize information. The year 2025 may witness these advancements.

2. Multimodal capabilities: Progressing from models that only support text to higher-level applications that accept image, video, or audio inputs.

3. Open-source AI: Open-source models are becoming valuable tools for companies to flexibly deploy and customize, and this trend is expected to expand further in the future.

4. Edge AI: AI that can run without a cloud connection, such as Apple Intelligence, protects data privacy through local execution and processing.

AI security trends

By 2025, as attackers employ more tactics, the Global cybercrime market is expected to expand. These cybercriminals will continue to exploit the instability of the geopolitical environment and extend cybersecurity challenges into new areas. However, innovations in AI agents will pave the way for the rise of multi-agent systems.

Multi-agent systems, where multiple autonomous agents collaboratively complete complex tasks, will become an important new field in AI security. Although these systems may face risks of data contamination, information injection, or social engineering attacks, they also provide new tools for cyber defense, such as early event response, application testing, and vulnerability detection.

In addition, Zero Trust strategy will further expand and enhance protection against internal threats. This strategy will integrate behavioral understanding dimensions to provide powerful Weapons for traditional security methods.

Conclusion

AI is not a passing trend; it is becoming a fundamental force of transformation in the Global business and economic landscape. Leaders who can devise strong AI strategy will be able to push their organizations to the forefront in terms of growth, innovation, and leadership.

With the rapid development of technology, particularly the rise of generative AI, there is little time left for companies that have yet to adopt AI-driven strategies to consider their options; failing to take action may pose significant risks of being eliminated.