We are living in 2025, and the question still lingers in many aspiring minds: is a deep, formal education in mathematics necessary to build a career in artificial intelligence? The answer is both yes and no. It has become more nuanced than ever before.
Let’s explore why.
The AI Landscape of the Past
If we rewind to 2018 or 2019, the world of AI was fundamentally different. Back then, practitioners were often building algorithms from the ground up. Researchers would devise new algorithmic approaches, and engineers would have to deeply understand and modify them to create a working model. This environment demanded a strong, hands-on command of mathematics. You were in the trenches, manipulating the core logic, which was all based on mathematical principles.
The Generative AI Revolution
So, what changed? The rise of Generative AI.
Today, we are no longer required to build every AI model from scratch. The paradigm has shifted dramatically. We now leverage powerful, pre-existing foundation models built by large research teams. Our job has evolved from being creators of algorithms to being architects who use these powerful tools. We now focus on fine-tuning, post-training, and post-processing these ready-made models to fit our specific business cases and applications.
This shift has significantly lowered the barrier to entry.
In her pivotal book, AI Engineering for Building Applications with Foundation Models, renowned AI researcher Chip Huyen notes that this revolution has democratized access to the field. Previously, a career in AI was gated by elite skills in math, programming, Python, and data science. That is no longer the absolute truth.
The Two Paths of an AI Engineer
Huyen’s book highlights a crucial distinction that has emerged in the field. The role of an “AI Engineer” has effectively split into two distinct paths:
-
The AI Research Engineer: This professional continues the traditional work of inventing and building algorithms from scratch. For this path, a profound understanding of mathematics remains non-negotiable. They are the ones pushing the theoretical boundaries of what is possible.
-
The AI Application Engineer: This is the role seeing a massive surge in demand across the industry. This engineer understands the fundamentals of how AI models work but doesn’t build them from zero. Instead, they apply existing models to create practical, real-world applications.
Think about it. ChatGPT is a sophisticated application built upon a foundation model (like GPT-3 or its successors) that was meticulously trained and fine-tuned. Similarly, incredible image generation tools like Midjourney are applications built upon powerful open-source models such as Stable Diffusion.
As an AI Application Engineer, your role is to take these powerful, well-crafted models and innovate on top of them. You avoid reinventing the wheel and instead focus on delivering value quickly and efficiently by leveraging the clean, robust work already done by major companies and research labs.
So, Do You Need Math in 2025?
Ultimately, the answer depends entirely on your career goals.
- If you aspire to be a researcher who creates the next generation of foundation models, then yes, a deep and rigorous study of mathematics is essential.
- If you want to be an application developer who builds innovative products and services using AI, then you can thrive by focusing on the practical skills of applying, fine-tuning, and integrating existing models.
The industry needs both, but the path for the application-focused engineer is now wider and more accessible than ever before.