AI Rediscovers Fundamental Physics Laws: NYU Abu Dhabi Study

The Groundbreaking NYU Abu Dhabi Study on AI and Particle Physics

In a remarkable advancement for both artificial intelligence and fundamental physics, researchers at New York University Abu Dhabi (NYUAD) have demonstrated that a relatively simple AI model can independently rediscover key principles of particle physics directly from raw experimental data. Published today in the prestigious Journal of High Energy Physics (JHEP), the study titled "Rediscovering the Standard Model with AI" showcases unsupervised machine learning techniques applied to historical particle data from the 1950s and 1960s. This work not only validates AI's potential in scientific discovery but also highlights NYU Abu Dhabi's pivotal role in UAE's burgeoning research ecosystem.

The Standard Model (SM) of particle physics is the prevailing theory describing three of the four fundamental forces (electromagnetic, weak, and strong nuclear forces) and classifying all known elementary particles, such as quarks, leptons, and bosons. Developed over decades through painstaking theoretical work and experiments like those at CERN's Large Hadron Collider (LHC), the SM underpins our understanding of matter and interactions at the subatomic scale. The NYUAD team's achievement lies in showing that AI, without any preconceived physics knowledge, can reconstruct these foundational elements from data alone.

Meet the Visionary Researchers Leading the Charge

Leading the study are Aya Abdelhaq, a researcher affiliated with both New York University and NYU Abu Dhabi; Pellegrino Piantadosi from NYU Abu Dhabi; and Fernando Quevedo from the University of Cambridge's Department of Applied Mathematics and Theoretical Physics (DAMTP). Abdelhaq and Piantadosi, based at NYUAD's Saadiyat Island campus, bring expertise in data science and physics, while Quevedo's contributions draw from his renowned work in string theory and quantum field theory.

NYU Abu Dhabi's Physics Division, part of its College of Arts and Sciences, fosters interdisciplinary research blending high-energy physics with computational methods. This study exemplifies how the university's global faculty—drawn from top institutions worldwide—drives innovation in the UAE. "This study shows that AI can uncover deep physical laws directly from data, opening the door to discovering new particles and patterns that traditional methods might miss," noted one of the researchers, underscoring the paradigm shift.

Step-by-Step: The AI Methodology Unveiled

The researchers fed the AI model with raw data on particle masses, spins, charges, and decay modes from mid-20th-century experiments—data that predates the quark model and full SM formulation. No equations for conservation laws or symmetry groups were provided; the AI operated purely on unsupervised learning.

• Dimensionality Reduction: Techniques like Principal Component Analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE) compressed high-dimensional data into visualizable clusters, revealing natural groupings of particles.
• Clustering Algorithms: Unsupervised clustering separated baryons (three-quark particles like protons) from mesons (quark-antiquark pairs), mirroring human classifications.
• Pattern Recognition: The model identified conserved quantum numbers such as baryon number (B=1 for baryons, 0 for mesons), isospin (SU(2) symmetry for up/down quarks), strangeness, charm, and even hints of bottom quark properties.

This process step-by-step replicated human discoveries: from the Eightfold Way (SU(3) flavor symmetry organizing particles into octets and decuplets) to Regge trajectories (linear mass-spin relations in particle families). The simplicity of the AI—no deep neural networks, just classical ML—makes it accessible and scalable.

Key Physics Principles Rediscovered by AI

The AI's outputs were stunningly accurate:

• Quantum Numbers and Symmetries: Precisely recovered baryon number, strangeness (S), charm (C), distinguishing particle families.
• Eightfold Way: Reproduced Gell-Mann and Zweig's classification, grouping hadrons into multiplets.
• Regge Trajectories: Plotted particle masses against spins, matching experimental Regge lines predictive of resonances.
• Interaction Strengths: Distinguished strong, electromagnetic, and weak decays via branching ratios.

These match decades of Nobel-winning work, from Murray Gell-Mann's quarks (1969) to the Higgs boson (2013). The preprint on arXiv (arXiv:2508.04923) details visualizations of these clusters.

Photo by Mathilde Cureau on Unsplash

Comparing AI Discovery to Human Physics Milestones

Human physicists like Enrico Fermi (1930s beta decay), Murray Gell-Mann (1960s quarks), and the CERN teams relied on theory guiding experiments. AI bypassed theory, letting data "speak." While humans took 30+ years post-1950s data to formalize the SM, AI did it in computation hours. This data-driven approach could spot anomalies in LHC data missed by model-biased analyses, accelerating beyond-SM physics like supersymmetry or dark matter candidates.

In UAE context, this aligns with national goals for AI-augmented research, positioning NYUAD alongside global leaders.

Implications for Particle Physics Research

Beyond validation, the study paves ways for:

• Discovering new particles/symmetries in LHC Run 3 data.
• Handling petabytes of collider data without human bias.
• Extending to cosmology (e.g., cosmic microwave background patterns) or quantum gravity.

Challenges remain: AI needs vast clean data; interpretability ("black box" risks) requires hybrid human-AI workflows.

NYU Abu Dhabi's Growing Leadership in UAE Higher Education

NYU Abu Dhabi, established in 2010 as NYU's portal campus, hosts 500+ global faculty driving research in physics, AI, and interdisciplinary fields. Its Center for Astrophysics and Space Science, Physics Division, and new MSc in Interdisciplinary Data Science and AI (MIDSAI) equip students for such breakthroughs. NYUAD contributes to UAE's Operation 300bn knowledge economy, with Abu Dhabi investing AED 100bn+ in R&D.

This study exemplifies UAE's shift from oil to innovation, fostering PhD programs and labs rivaling MIT/Caltech.

AI's Expanding Role in UAE Universities

UAE leads MENA in AI: Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) focuses purely on AI; Khalifa University advances quantum/AI; Masdar Institute tackles energy physics. NYUAD collaborates via events like EMNLP. UAE's National AI Strategy 2031 aims for 20% GDP from AI by 2031, funding 1000+ AI specialists annually. Physics depts integrate ML for simulations, aligning with Vision 2031.

Photo by Sajimon Sahadevan on Unsplash

Impacts on UAE's Economy, Jobs, and Future Talent

This research boosts UAE's R&D: AI-physics fusion spurs startups in quantum computing, semiconductors. Jobs surge in research assistants, postdocs, faculty at NYUAD/KU/MBZUAI. UAE universities offer 5000+ AI/physics positions yearly, with salaries AED 20k-50k/month for PhDs. Students gain hands-on via MIDSAI, preparing for global careers.

Future Outlook: AI-Driven Discoveries in UAE

Next: Apply to LHC data for new physics; integrate with UAE's quantum initiatives. NYUAD plans AI-physics labs, collaborations with CERN. For UAE HE, expect curriculum shifts: ML mandatory in physics BSc/MSc. Actionable: Aspiring researchers, pursue NYUAD/KU PhDs; explore research jobs in UAE.

This NYUAD milestone cements UAE as AI-research hub, promising transformative science.