Copy of Artificial Intelligence for Engineering Sciences

Background - Jensen Huang and emphasis on Physical Sciences

Since the first time I heard about the idea from Jensen Huang about focussing on Physical Sciences , I have been fascinated by it - both from an AI perspective - but also from a skills perspective

From a broader evolution of AI perspective, this makes sense since it aligns to the emphasis on world models see this article from Quanta - ‘World Models,’ an Old Idea in AI, Mount a Comeback

A world model is a representation of the environment that an AI carries around inside itself.  The AI system can use this simplified representation to evaluate predictions and decisions before applying them to its real-world tasks.

When Jensen Huang talks about focusing on the physical sciences, he isn’t pointing to a single subject — he’s talking about returning to the roots of how the world works. He’s urging future engineers to study physics, chemistry, materials science, mechanics, and earth or space sciences — disciplines that explain the laws governing matter, motion, energy, and force.

In my view, that certainly aligns to the world model of AI.

His message is clear: as AI evolves, the next breakthroughs won’t come from writing code alone. They’ll come from understanding the physical world deeply enough to model, predict, and interact with it. We’re moving from systems that see and generate (like today’s generative AI) toward systems that reason and act — what Huang calls “Reasoning AI” and eventually “Physical AI.”

This new kind of AI will need to grasp how real objects move, how materials behave, and how forces interact — the domain  of physics and mechanics. It’s about building machines that don’t just process data but can think within the constraints of the real world: predict the impact of motion, understand cause and effect, manipulate physical objects, and navigate environments.

In essence, Huang is calling for a fusion — between AI and the physical sciences — where knowledge of mechanics, materials, and natural systems becomes as crucial to building intelligent machines as computer science itself.

Engineering Sciences

I very much agree to this vision - and it aligns with the world view aspect of AI - especially through the NVIDIA Nvidia Omniverse

I have been working on the idea of Artificial Intelligence for Engineering Sciences.

Engineering Sciences form the scientific bedrock of engineering practice — the shared body of knowledge that links mathematics, physics, chemistry, biology, and computer science to how we design, analyze, and optimize systems, materials, and processes. Engineering Sciences occupies the middle ground between pure science and applied engineering, turning natural laws into principles that engineers can use to solve practical problems and create innovation.

At its core, Engineering Science is the systematic study of the physical, chemical, biological, and computational principles that give structure to every branch of engineering. It provides the tools to predict, model, and control how both natural and human-made systems behave.

It is analytical at its foundation. The need to understand physical systems involves working with mathematical modeling, thermodynamics, mechanics, electromagnetism, materials science, and systems theory. It has a predictive purpose, focused on explaining how systems behave under established physical laws like the Navier–Stokes equations, Hooke’s law, or Maxwell’s equations. 

Engineering Sciences serves as a bridge—linking the “why” of scientific theory with the “how” of engineering design. It is interdisciplinary by nature, weaving together the methods and insights of physics, chemistry, biology, and computation to create frameworks that span mechanical, civil, electrical, chemical, biomedical, and emerging fields such as AI and quantum engineering. Above all, it is quantitative and model-centric, using analytical and computational models to extract insight long before a single prototype is built or an experiment is run.

Across the disciplines, engineering science takes different forms:

• In mechanical engineering, it’s seen in continuum mechanics, fluid dynamics, and thermodynamics.
• In electrical engineering, it emerges through electromagnetic theory, circuit theory, and control systems.
• In chemical engineering, it underpins reaction kinetics, transport phenomena, and process design.
• In civil engineering, it supports structural mechanics, geomechanics, and materials science.
• In computer and AI engineering, it manifests in information theory, optimization, and systems modeling.

In the twenty-first century, AI and computation have become a new layer of engineering science. With data-driven models, simulation, and causal reasoning, engineers can now explore the behavior of complex systems in ways that were once impossible. This emerging domain—AI for Engineering Sciences—unites scientific understanding and engineering design within a single framework, laying the groundwork for a new generation of cognitive, intelligent, and autonomous systems.

Building next generation skills in AI for Engineering Sciences

Based on the above idea, the next question is: How do we build skills in AI for Engineering Sciences for the next generation? A lot is at stake here - innovation, new services and competitive advantage of nations. 

To develop the next generation skills in AI for Engineering Sciences for the next generation - I believe we should approach this in three parts

1. Part One: Foundations: What Engineers Should Understand About AI in the Engineering Sciences
2. Part Two: Present: How AI Is currently being used across Engineering Sciences
3. Part Three: Future: Where AI and Engineering Sciences Are Headed

This provides a much deeper foundation than merely listing areas where machine learning and deep learning are used in engineering. More importantly. It inculcates the mindset of an AI researcher for Engineering sciences 

Part One - Foundations: What Engineers Should Understand About AI in the Engineering Sciences

The synergy  between artificial intelligence and the engineering sciences involves the meeting of computational intelligence, scientific modeling, and design thinking - as one unified discipline. For engineers entering this space, we start with understanding the mathematics and concepts that power AI—linear algebra, calculus, optimization, probability, and statistics. 

They are the same principles used in control systems, mechanics, and simulations - but these are  now expanded to drive algorithms like neural networks, support vector machines, and probabilistic models. 

Engineers also need to be fluent in the main families of machine learning—supervised, unsupervised, and reinforcement learning—each offering new ways to extend traditional system modeling into data-rich, adaptive environments.

AI’s growing role in modeling and simulation has created hybrids that combine theoretical and empirical thinking. Physics-informed neural networks (PINNs) embed governing equations directly into learning systems. Surrogate models approximate expensive CFD or FEM simulations, dramatically speeding up design loops. Digital twins close the feedback loop by merging live sensor data with predictive models to simulate and test systems in real time. Add to that the creative potential of generative models—VAEs, GANs, diffusion networks—which can propose new geometries or materials under physical constraints.

In control systems, AI builds on classical theory with reinforcement learning and multi-agent coordination, enabling adaptive behavior in environments that change constantly. Understanding cyber-physical systems—where computation, sensing, and actuation intertwine—is essential. Engineers must also learn the language of reasoning: causal graphs, knowledge representations, and neuro-symbolic systems that bring together logic and learning. These tools don’t just improve performance; they make systems explainable and trustworthy, vital traits in safety-critical engineering.

Beyond algorithms, engineers should master data pipelines, MLOps practices, and uncertainty quantification—the foundations that make AI systems reliable. And there’s a philosophical layer too: understanding what it means for an AI system to “know,” to reason under uncertainty, and to make decisions alongside humans. These questions bring ethics and epistemology into the technical fold. 

Altogether, these foundations prepare engineers not just to use AI, but to think with it—to move fluidly between physical law, data, and cognition.

Present: How AI Is currently being used across Engineering Sciences

AI is already incorporated into  engineering processes. It shapes how we design, build, and operate almost every kind of system. At its core, the shift is about optimization through data—using AI to improve efficiency, reliability, and innovation.

 In design, AI accelerates exploration with generative design tools and topology optimization, while evolutionary algorithms help engineers test countless possibilities in silico. Across sectors like aerospace, automotive, and civil engineering, surrogate models make it possible to evaluate complex systems in seconds rather than hours.

In materials and manufacturing, AI has become an engine of discovery. Materials informatics predicts the properties of new compounds and alloys long before they’re synthesized. Machine learning keeps production lines running smoothly, powering predictive maintenance and process optimization. In additive manufacturing, neural networks monitor prints layer by layer, adjusting parameters in real time to prevent defects. 

The energy sector leans heavily on AI for forecasting, grid management, and renewable integration—turning the dream of flexible, sustainable energy systems into something more tangible.

AI’s influence reaches deep into mechanical, aerospace, and civil engineering through smarter simulations and monitoring systems. Machine learning models stand in for expensive CFD or FEM runs, while AI-based structural health monitoring spots fatigue or cracks that human inspectors might miss. In cities, AI supports urban planning, improving traffic flow, managing building energy, and modeling disaster risks from earthquakes or floods.

The chemical and process industries rely on AI for reaction prediction, fault detection, and process control. In electrical and electronic engineering, AI drives smarter signal processing, predictive control, and fault diagnosis in power grids and devices. The biomedical field merges AI with biomechanics and imaging to advance prosthetics and digital replicas of the human body.

At the systems level, AI underpins cyber-physical systems, robotics, and automation, fusing sensing, reasoning, and action. It also keeps supply chains running efficiently, predicting disruptions and optimizing logistics. Meanwhile, in research settings, AI-assisted discovery is becoming the norm: symbolic regression unearths governing equations, and “self-driving labs” run experiments without human oversight. 

Thus, AI is no longer a peripheral tool. It’s already an intrinsic part of engineering itself, expanding how we explore, validate, and design.

Future: Where AI and Engineering Sciences Are Headed

The next era of engineering won’t just use AI—it will collaborate with it. 

AI is poised to move from predictive modeling to genuine reasoning and co-creation. The rise of cognitive engineering systems is a sign of that shift. Tomorrow’s digital twins won’t just simulate—they’ll reason, explain, and propose new possibilities. Engineers will interact with these models as conversational partners, not black boxes, asking them why a design works, not just how well.

Another frontier lies in autonomous design and co-creation. Imagine AI agents generating and testing new designs continuously, guided by generative physics engines that learn the underlying patterns of nature. In these shared human–AI workspaces, creativity becomes collaborative: engineers guide objectives and ethics; AI explores the vast design landscape. This blurs the line between design and discovery.

AI will also become an active player in scientific exploration—uncovering laws, relationships, and mechanisms hidden in data. Symbolic regression and autonomous labs will make it possible for AI to propose theories, run experiments, and refine its own understanding. In multiscale and multiphysics modeling, AI will knit together simulations that span quantum, molecular, and continuum levels—producing unified solvers far faster and more interpretable than current numerical approaches.

On a larger scale, AI will orchestrate systems-of-systems—energy grids, transportation networks, and water systems—through adaptive, multi-agent collaboration. These interconnected systems will predict disruptions, reconfigure themselves, and maintain resilience autonomously. Sustainability will become a core engineering principle: carbon-aware optimization, circular design, and regenerative infrastructures will all be guided by intelligent models that understand feedback loops between the built and natural worlds.

The human side of engineering will change too. Future engineers will work as part of human–AI design teams, learning to interpret and guide AI’s reasoning rather than merely consuming its outputs. Education will focus as much on metacognition—thinking about thinking—as on technical skill. New frameworks like AI-driven compliance systems and epistemic AI will help ensure trust and accountability in complex systems.

Looking further ahead, AI will expand beyond digital computing. Quantum machine learning could tackle problems once thought unsolvable, while neuromorphic and biohybrid systems will bring adaptability and energy efficiency to embedded control. The horizon points to self-evolving engineering ecosystems—recursive systems capable of improving their own models, designs, and reasoning.

If the last century was about building machines that extend human power, the next will be about creating systems that extend human thought. 

Conclusion

The future of AI in engineering rests on three directions: Cognitive—AI that reasons and discovers; Autonomous—AI that learns and self-designs; and Sustainable—AI that aligns with the rhythms of society and the planet. Together, they signal a profound shift: from engineering as control over the physical world to engineering as a partnership with intelligence itself.

As you can see, I am taking this idea from Jensen Huang very seriously in the future of skills in AI and Engineering Sciences

This post took a long time to write - because I changed perspectives - finally focusing on the idea of going from traditional engineering to AI researcher in engineering sciences - which I believe is an emerging discipline. 

Comments welcome especially if you are working in this space.