Rafael Gómez-Bombarelli says we’re at science’s ‘second inflection’—AI reasoning over text, structures, and recipes to invent materials.
What if AI didn’t just analyze data, but invented new plastics and batteries from scratch? MIT’s Rafael Gómez-Bombarelli calls this the dawn of general scientific intelligence via multimodal LLMs.[1][7]
His lab blends physics sims, ML, and gen AI for fully computational workflows: high-throughput data generation improves models, which prioritize experiments for partners in OLEDs, catalysts, and more. He’s spun out companies to commercialize.[1][7]
For devs: build tools that chain sims + VLMs for end-to-end discovery—prototype materials in hours, not years. Accelerate your chem-eng, energy, or pharma stacks with ‘AI chemists’.[1]
Beats siloed tools like AlphaFold (proteins-only) by reasoning multimodally over recipes/structures/text, aligning with JAIST’s knowledge fusion for broader domains. Science’s ChatGPT moment.[1][2]
Fork his open tools on GitHub, pair with PySCF for sims—ready to design your first AI catalyst? Watch for startup acquisitions; this shifts science from hypothesis to hyperscale.[1][7]
Source: NeuralBuddies & MIT News