[Conference] 🏙️ AI + Building Simulation Conference Hong Kong 2026

How is artificial intelligence actually transforming buildings and cities?

At Building Simulation 2026 (BAS 2026) in Hong Kong, hosted at HKUST, I spent three days looking beyond the AI hype to understand how these methods are actually being applied in building science, energy systems, and climate-neutral cities. The conference itself placed climate-neutral buildings and cities at its core, with AI appearing across keynotes, technical sessions, and workshops.

What struck me was the diversity of AI approaches now entering our field:

🤖 Large Language Models (LLMs) for automating building energy simulation, EnergyPlus workflows, optimization, anomaly detection, and control

🌡️ Computer vision for extracting information from infrared, radar, and real-building data

🏙️ Graph Neural Networks (GNNs) for urban-scale building energy prediction

⚙️ Physics-Informed Neural Networks (PINNs) and surrogate models for reducing the computational burden of conventional simulation

🎛️ Reinforcement Learning (RL) for HVAC operation, demand response, and real-time control

📊 Generative AI for probabilistic building energy forecasting

🏢 Digital twins and AI-driven energy management

⚡ Smart grids, energy flexibility, and data centers

These are no longer isolated topics. The conference program explicitly connected LLMs and data-driven modeling with building-grid interaction, smart grids, microgrids, urban energy modeling, and climate-neutral districts.

One of the most interesting shifts was the growing connection between AI, computing infrastructure, and building energy use.

AI requires computing power. Computing requires data centers. Data centers require enormous amounts of electricity and cooling.

On the final day, this connection became explicit through discussions of energy flexibility in buildings and data centers, followed by digital twins and AI-driven energy management for grid-interactive buildings.

This raises a question I think our research community needs to take seriously:

Can AI help us decarbonize buildings without creating a new energy and carbon problem of its own?

That means looking behind model accuracy and asking harder methodological questions: Is the work reproducible? What are the GPU and CPU requirements? What does training cost? Has the model been validated against measured data? Can it actually be deployed in a real building? And what is the carbon footprint of the computation itself?

My main takeaway for students and researchers is simple:

👉 The most sophisticated AI model is not necessarily the best model.

A well-validated surrogate model that solves a meaningful building problem, runs efficiently, and can be deployed with modest computational resources may ultimately be far more valuable than a highly complex model requiring massive computing infrastructure. This was also the central reflection I took from the conference: focus on measured-data validation, deployment, reproducibility, and the minimum necessary complexity.

The future of AI in building science should therefore not only be about making our models smarter. It should be about making them useful, transparent, reproducible, computationally responsible, and capable of delivering measurable decarbonization in the real world.

💬 For researchers working with AI and building simulations: where do you see the greatest real-world potential, and where do you think we are still seeing more hype than impact?

Dejan Mumovic | Qingyan Chen | Jerry Yan | Jianlei Niu | Pieter de Wilde | Alfonso Capozzoli | Da Yan | Borong Lin | Liangzhu Leon | Marco Savino Piscitelli Shikang Weng

I look forward to what we may build together. Check the conference program: https://www.bas2026.org/Program.html

📽️ Watch the academic vlog about the visit 👉 YouTube: AI + Buildings at Building Simulation 2026 https://www.youtube.com/watch?v=uQE7_Hcbjo0

📚 Learn more about our research: https://www.sbd.uliege.be/
💻 Subscribe to my newsletter: https://lnkd.in/diTVT5eq
🅱️ Bilibili: b23.tv/bzjL3bn | 🎬 YouTube: https://lnkd.in/erHrfkNf
🌐 Explore previous posts and resources: https://www.shadyattia.org/
🔗 Follow all my professional links, including WeChat 🟩💬 微信: https://lnkd.in/eN3xZhhZ

#BuildingSimulation2026 #BAS2026 #ArtificialIntelligence #BuildingSimulation #BuildingScience #LLM #EnergyPlus #MachineLearning #DigitalTwins #PINNs #GraphNeuralNetworks #ReinforcementLearning #HVAC #SmartGrids #DataCenters #BuildingEnergy #ClimateNeutralBuildings #SustainableCities #EnergyFlexibility #Decarbonization #HKUST #HongKong #SBDLab