AIght_
ToolsLearnFieldsUniverseSignalHumanAbout
Take the quiz
← All fields

Field guide

Physics & Engineering

AI is compressing simulation timelines from weeks to seconds, enabling physics-informed design of complex systems from fusion reactors to next-generation materials.

Medium
Start your path →See your personal disruption score in 2 minutes

What's changing

01

Physics-informed neural networks and surrogate models now simulate fluid dynamics, quantum states, and structural performance up to 1,000x faster than traditional solvers, allowing real-time design iteration.

02

AI agents analyze LHC particle collision data to detect rare events amid billions of mundane ones and propose simplifying transformations in theoretical physics equations.

03

Generative AI tools optimize ECU calibration, battery chemistry, and engineering workflows, reducing physical testing needs while maintaining accuracy in areas like aerospace and automotive design.

A surrogate model isn't a replacement for a real simulation — it's a way to skip 90% of them and only run the ones that matter.

A path through the universe

How to actually learn AI for Physics & Engineering.

Two tracks. Pick your depth. The left one gets you fluent for conversations and tool choices. The right one is what you read when you actually want to know how it works.

Intuitions

No math required.

  1. 01Chain-of-ThoughtWhen 'think step by step' actually earns its keep — and when it's just expensive theater.6 min read
  2. 02Hallucination & GroundingWhy AI models confidently make things up — and what you can actually do about it8 min read
  3. 03Multimodal ModelsWhen AI learned to see, listen, and read — at the same time, in the same head7 min read
  4. 04Prompt EngineeringThe craft of talking to a model that will take you exactly as literally as it decides to7 min read
  5. 05Structured OutputForcing the model to fill in a shape — and why it's harder than it looks.5 min read

Goes deeper

Under the hood.

  1. 01TransformersThe architecture that changed what AI could do with language — and then everything else8 min read
  2. 02How AI Models Are TrainedFrom random noise to a model that can reason — the actual pipeline10 min read
  3. 03Scaling LawsWhy bigger keeps working — and the question of where it stops.7 min read
  4. 04Function CallingThe JSON-shaped API that turned chat models into clients of the real world.6 min read
  5. 05Fine-TuningTeaching a model new habits, not new knowledge8 min read

AI impact spectrum

Automated

  • Standard simulations
  • CAD drafting
  • Report generation

Augmented

  • Physics-informed modeling
  • Materials optimization
  • Multi-variable design

Growing

  • Systems thinking
  • Novel physics research
  • Safety/ethics oversight

Roles at risk

CAD draftsman

Routine simulation analyst

Manual test & measurement technician

Roles growing

Physics-AI hybrid engineer

Generative design specialist

Digital twin architect

AI validation engineer

CAD draftsmen aren't being replaced by AI. They're being replaced by AI-equipped designers who used to be CAD draftsmen.

What to actually do

Over the next two years, physicists and engineers should embed AI into simulation pipelines by training or fine-tuning physics-informed models on their domain-specific datasets, replacing lengthy finite-element runs with surrogate AI models for rapid prototyping, and using tools like Neural Concept to explore 10x more design variants before physical builds. Actively validate AI outputs against known physics benchmarks, contribute domain data to open models, and shift focus from manual computation to interpreting AI-generated insights—while learning to prompt multimodal LLMs for hypothesis generation in theoretical work.

If you're validating an AI surrogate, hold out a non-trivial benchmark from training. Lots of papers don't, and lots of surrogates lie.

Sources

  1. [1]Raissi et al., Physics-informed neural networks (2019)
  2. [2]Karniadakis et al., Physics-informed machine learning (2021)
Medium

Engineers already use simulation software, so AI extensions build on existing skills, but incorporating physics constraints and validation requires targeted upskilling.

Personalize this

How disrupted are you, really?

Three questions. An honest score tailored to your specific role.

Take the quiz →

Tools to know

Neural Concept

Uses deep learning for physics simulations (CFD/FEA) to predict design performance faster than traditional methods

Ansys AI

Accelerates structural, thermal, and fluid simulations with integrated AI for real-time feedback

Altair PhysicsAI

Generates reduced-order models for complex physics outcomes using geometric deep learning

Concepts to understand

Physics-informed neural networks (PINNs)Surrogate modeling for high-fidelity simulationsGenerative design with physics constraints

Get your personal disruption score

Based on your specific role within Physics & Engineering

Run AI impact quiz →Explore other fields