Source: Quantum Magazine
As I was writing this article, I thought, rather than writing "what to expect", it would be simpler to write "what not to expect from AI in 2025"! For a technology that is making progress simultaneously in almost every field, how can one choose 5 areas? To narrow the scope and hopefully make it more interesting, I decided to choose trends that have no direct connection to the development of ChatGPT or its competitors. It can be said with certainty that these trends will grow, and these applications and the companies behind them will strive to make these applications the solution to all possible problems.
Here are 5 AI trends for 2025 (in no particular order):
Trend 1: Agents Everywhere
You may have heard the term "Agentic AI". While AI has always been about learning patterns, AI has evolved to the following stages: (a) learning patterns from data (b) generating new content based on these patterns, and (c) taking actions based on these patterns. When these three come together, you have an AI agent - a software that can learn, create actions, and execute those actions. This area is expected to see more development in 2025.
Trend 2: Transformation of the Education System
There has been much discussion about whether AI will encourage cheating, replace teachers, or fundamentally change the way students learn in other ways. While all of these are crucial, another equally important, if not more so, force is emerging. This year, there is increasing evidence that graduating students are unable to find jobs due to AI-driven skills and economic conditions. This raises not only the question of how students should learn, but also what they should learn. The economic pressure from the downturn in the job market will force graduating students, and ultimately the institutions that educate them, to face the new demands of employers for employees. Students will need to adapt first, upskilling through various means, and institutions will need to keep up. I expect we will start to see these changes in 2025.
Trend 3: Application of AI in Science
Two of the 2024 Nobel Prizes in Science were related to AI. This should sound the alarm: the application of AI in science will continue. Also noteworthy is that while the world's attention and imagination are focused on generative AI, billions of dollars are pouring into the scientific applications of AI, from space exploration to medical advances, with new announcements every day. It is also worth noting that despite all this investment and progress, data shows that AI-discovered drugs have about the same success rate in Phase 2 clinical trials as other drugs, but some of these have already been "known" in some form. As of the writing of this article, I have not seen any news of an AI-generated drug being approved by the US Food and Drug Administration. What does this combination tell us? It tells us that the potential is huge, but has not yet been fully realized.
Trend 4: Data Mining
Skeptics have long predicted that AI will exhaust data, while others have refuted this. In these predictions, the consistent thing is not the existence of data, but the increasing difficulty of obtaining high-quality and ethically compliant data. I expect this to be a trend in 2025. Untapped data, especially on our physical environment, is still immense. However, large language models have already scoured most of the easily accessible data. In 2025, it is expected that more and more efforts will be made, whether through commercial contracts to obtain data, through labeling systems to organize undeveloped data, or through the deployment of more sensors, and so on. Combined with the aforementioned trend of AI in science, we can imagine that the effort to mine scientific data will accelerate.
Trend 5: Robotics
AI has made progress in all areas where problems can be solved through software (such as email, content creation, MRI analysis, etc.). In all these areas, AI has driven cost savings and job disruption. Robotics technology brings AI into the physical realm - whether in manufacturing, surgery, agriculture, or space exploration. The applications of the combination of AI and physical automation are almost endless. In 2025, we can expect to see the existing trends in this field continue to expand and attract wider public attention.
In Conclusion
In the past year, large language models and generative AI have seen rapid development, seemingly able to solve any basic task. In 2015, as the focus shifts to deeper impacts on specific domains and institutions, as well as the integration with other technological waves, we can expect to see the next wave arrive.
