General Purpose Models for the Chemical Sciences ✨

📘 Read the Review

You can read the review directly on or access it as a collaborative online book at:

👉 arXiv

👉 Online Book

We also welcome community contributions — this is a living resource of references we have used in the review, and we aim to keep current with the evolving ecosystem. If you don't find your work or your favourite work here, please add it.

✨ Join the Community

Help us grow and improve this resource by sharing your feedback or contributing directly via the online platform.

Together, we can keep this reference useful, relevant, and up to date for everyone! This document provides an overview of the research sections and their associated references.

Introduction

• 2025 - From text to insight: large language models for chemical data extraction
• 2025 - Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning
• 2025 - System Card: Claude Opus 4 & Claude Sonnet 4
• 2024 - A holistic platform for accelerating sorbent-based carbon capture
• 2024 - Fine-tuning large language models for chemical text mining
• 2024 - Structured information extraction from scientific text with large language models
• 2024 - Leveraging large language models for predictive chemistry
• 2024 - Automatic Prediction of Molecular Properties Using Substructure Vector Embeddings within a Feature Selection Workflow
• 2024 - Inverse design workflow discovers hole-transport materials tailored for perovskite solar cells
• 2024 - Unraveling Molecular Structure: A Multimodal Spectroscopic Dataset for Chemistry
• 2024 - Elucidating Structures from Spectra Using Multimodal Embeddings and Discrete Optimization
• 2024 - Prospective de novo drug design with deep interactome learning
• 2024 - Probing the limitations of multimodal language models for chemistry and materials research
• 2024 - t-SMILES: a fragment-based molecular representation framework for de novo ligand design
• 2024 - Invalid SMILES are beneficial rather than detrimental to chemical language models
• 2024 - Chemllm: A chemical large language model
• 2024 - nach0: Multimodal natural and chemical languages foundation model
• 2024 - OneProt: Towards Multi-Modal Protein Foundation Models
• 2023 - Scientific discovery in the age of artificial intelligence
• 2023 - Autonomous chemical research with large language models
• 2023 - Machine learning for industrial processes: Forecasting amine emissions from a carbon capture plant
• 2023 - GPT-4 Technical Report
• 2023 - A foundation model for atomistic materials chemistry
• 2023 - The future of chemistry is language
• 2022 - The case for data science in experimental chemistry: examples and recommendations
• 2022 - Machine learning for a sustainable energy future
• 2022 - Making the collective knowledge of chemistry open and machine actionable
• 2022 - Chemberta-2: Towards chemical foundation models
• 2022 - A universal graph deep learning interatomic potential for the periodic table
• 2021 - Origins of structural and electronic transitions in disordered silicon
• 2021 - Combining Machine Learning and Computational Chemistry for Predictive Insights Into Chemical Systems
• 2021 - Using automated serendipity to discover how trace water promotes and inhibits lead halide perovskite crystal formation
• 2021 - Machine learning force fields
• 2020 - Big-data science in porous materials: materials genomics and machine learning
• 2020 - When machine learning meets multiscale modeling in chemical reactions
• 2020 - Language models are few-shot learners
• 2019 - A robotic platform for flow synthesis of organic compounds informed by AI planning
• 2018 - Machine learning for molecular and materials science
• 2017 - Use machine learning to find energy materials
• 2014 - The significance of implicit knowledge for learning and teaching chemistry
• 2012 - Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning
• 2009 - The tacit dimension

The Shape and Structure of Chemical Data

Shape of Scientific Data

• 2025 - Deep Learning is Not So Mysterious or Different
• 2022 - Bridging known and unknown unknowns: From natural products and their mimics to unmet needs in neuroscience
• 2021 - Belousov--Zhabotinsky type reactions: the non-linear behavior of chemical systems
• 2021 - Machine learning force fields
• 2021 - Physics-Inspired Structural Representations for Molecules and Materials
• 2017 - Introduction: Ten Theses on Big Data and Computability

Scale of Chemical Data

• 2025 - ChemPile: A 250GB Diverse and Curated Dataset for Chemical Foundation Models
• 2024 - The Llama 3 Herd of Models
• 2023 - NOMAD: A distributed web-based platform for managing materials science research data
• 2021 - The Open Reaction Database
• 2019 - Beware of plausible predictions of fantasy materials
• 2019 - Hydrogen Storage Materials Database
• 2014 - Quantum chemistry structures and properties of 134 kilo molecules
• 2008 - Shedding light on the dark data in the long tail of science
• 2002 - Molecular biology of the cell

Dataset Creation

• 2025 - Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning
• 2024 - Common Crawl
• 2024 - The fineweb datasets: Decanting the web for the finest text data at scale
• 2024 - Rephrasing the Web: A Recipe for Compute and Data-Efficient Language Modeling
• 2024 - Rephrasing natural text data with different languages and quality levels for Large Language Model pre-training
• 2023 - The refinedweb dataset for falcon llm: Outperforming curated corpora with web data only

Filtering

• 2025 - ChemPile: A 250GB Diverse and Curated Dataset for Chemical Foundation Models
• 2023 - Textbooks Are All You Need
• 2023 - When Less is More: Investigating Data Pruning for Pretraining LLMs at Scale
• 2023 - CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society
• 2020 - Scaling Laws for Neural Language Models
• 2020 - The Hardware Lottery
• 2019 - Value-laden Disciplinary Shifts in Machine Learning
• 2017 - Enriching word vectors with subword information
• 2012 - Imagenet classification with deep convolutional neural networks

Synthetic Data

• 2025 - ChemPile: A 250GB Diverse and Curated Dataset for Chemical Foundation Models
• 2025 - Assessment of fine-tuned large language models for real-world chemistry and material science applications
• 2024 - Leveraging large language models for predictive chemistry
• 2024 - Evaluating Chemistry Prompts for Large-Language Model Fine-Tuning
• 2024 - Rephrasing the Web: A Recipe for Compute and Data-Efficient Language Modeling
• 2024 - MathGenie: Generating Synthetic Data with Question Back-translation for Enhancing Mathematical Reasoning of LLMs
• 2024 - CantonMT: Cantonese to English NMT Platform with Fine-Tuned Models Using Synthetic Back-Translation Data
• 2024 - Deepseek-prover: Advancing theorem proving in llms through large-scale synthetic data
• 2024 - Solving olympiad geometry without human demonstrations
• 2024 - Collapse or Thrive? Perils and Promises of Synthetic Data in a Self-Generating World
• 2024 - AI models collapse when trained on recursively generated data
• 2023 - Darwin series: Domain specific large language models for natural science
• 2023 - Chemical representation learning for toxicity prediction
• 2022 - RanDepict: Random chemical structure depiction generator
• 2022 - Application of data augmentation techniques towards metabolomics
• 2022 - Autoformalization with Large Language Models
• 2021 - Text data augmentation for deep learning
• 2021 - Maxsmi: maximizing molecular property prediction performance with confidence estimation using smiles augmentation and deep learning
• 2020 - General protocol for the accurate prediction of molecular 13C/1H NMR chemical shifts via machine learning augmented DFT
• 2019 - A survey on image data augmentation for deep learning
• 2019 - EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks
• 2019 - Randomized SMILES strings improve the quality of molecular generative models
• 2019 - Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks
• 2018 - A convolutional neural network-based screening tool for X-ray serial crystallography
• 2018 - Contextual Augmentation: Data Augmentation by Words with Paradigmatic Relations
• 2018 - Understanding Back-Translation at Scale
• 2015 - The Lean theorem prover (system description)
• 2008 - The isabelle framework

Building Principles of GPMs

Taxonomy of Foundation Models

• 2020 - Language models are few-shot learners
• 2020 - Retrieval-augmented generation for knowledge-intensive nlp tasks

Representations

• 2023 - What algorithms can transformers learn? a study in length generalization
• 2022 - Assessing the Impact of Sequence Length Learning on Classification Tasks for Transformer Encoder Models
• 2018 - Comment on “Predicting reaction performance in C--N cross-coupling using machine learning”

Common Representations of Molecules and Materials

• 2024 - Designing proteins with language models
• 2023 - Group SELFIES: a robust fragment-based molecular string representation
• 2022 - ProtTrans: Toward Understanding the Language of Life Through Self-Supervised Learning
• 2022 - What Information is Necessary and Sufficient to Predict Materials Properties using Machine Learning?
• 2022 - Representations of molecules and materials for interpolation of quantum-mechanical simulations via machine learning
• 2022 - E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials
• 2021 - Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences
• 2021 - Compositionally restricted attention-based network for materials property predictions
• 2021 - Physics-Inspired Structural Representations for Molecules and Materials
• 2021 - E (n) equivariant graph neural networks
• 2020 - Self-referencing embedded strings (SELFIES): A 100% robust molecular string representation
• 2019 - Identification Schemes for Metal–Organic Frameworks To Enable Rapid Search and Cheminformatics Analysis
• 2018 - ElemNet: Deep Learning the Chemistry of Materials From Only Elemental Composition
• 1991 - The crystallographic information file (CIF): a new standard archive file for crystallography
• 1988 - SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules

Tokenization

• 2022 - Improving graph neural network expressivity via subgraph isomorphism counting

Embeddings

• 2019 - Unsupervised word embeddings capture latent knowledge from materials science literature
• 2013 - Efficient Estimation of Word Representations in Vector Space
• 2013 - Distributed Representations of Words and Phrases and their Compositionality

General Training Workflow

• 2018 - Universal language model fine-tuning for text classification

Pre-training: Learning the Shape of Data

• 2020 - Language models are few-shot learners

Self-Supervision

Families of Self-Supervised Learning

Generative Methods

• 2025 - Scientific large language models: A survey on biological & chemical domains
• 2024 - MatText: Do language models need more than text & scale for materials modeling?
• 2024 - Pre-training with fractional denoising to enhance molecular property prediction
• 2023 - cMolGPT: a conditional generative pre-trained transformer for target-specific de novo molecular generation
• 2023 - Denoise pretraining on nonequilibrium molecules for accurate and transferable neural potentials
• 2022 - Molecular contrastive learning of representations via graph neural networks
• 2022 - Graph neural networks for materials science and chemistry
• 2021 - Masked graph modeling for molecule generation
• 2021 - Generative Pre-Training from Molecules
• 2020 - {ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Prediction}
• 2019 - Molecular transformer: a model for uncertainty-calibrated chemical reaction prediction
• 2018 - BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
• 2013 - Generalized denoising auto-encoders as generative models
• 2010 - Stacked denoising autoencoders: Learning useful representations in a deep network with a local denoising criterion
• 2008 - Extracting and composing robust features with denoising autoencoders

Contrastive Learning

• 2018 - Representation Learning with Contrastive Predictive Coding
• 2018 - Deep Clustering for Unsupervised Learning of Visual Features
• 2006 - Dimensionality reduction by learning an invariant mapping

The Holy Grail of Building Good Internal Representation

• 2025 - UMA: A Family of Universal Models for Atoms
• 2025 - Accurate predictions on small data with a tabular foundation model
• 2023 - ImageBind: One Embedding Space To Bind Them All
• 2022 - Generalized Visual Language Models
• 2022 - MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields

Fine-Tuning: Learning the Coloring of Data

• 2024 - Leveraging large language models for predictive chemistry

Post-Supervised Adaptation: Learning to Align and Shape Behavior

• 2025 - Does Math Reasoning Improve General LLM Capabilities? Understanding Transferability of LLM Reasoning
• 2025 - Training a Scientific Reasoning Model for Chemistry
• 2025 - Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Models
• 2022 - Training language models to follow instructions with human feedback
• 2017 - Proximal Policy Optimization Algorithms
• 2015 - Sample complexity of episodic fixed-horizon reinforcement learning

Example Architectures

• 2025 - Bio-xLSTM: Generative modeling, representation and in-context learning of biological and chemical sequences
• 2025 - Transformers are Graph Neural Networks
• 2024 - MatText: Do language models need more than text & scale for materials modeling?
• 2023 - Everything is connected: Graph neural networks
• 2023 - Mamba: Linear-Time Sequence Modeling with Selective State Spaces
• 2017 - Attention Is All You Need
• 1997 - Long short-term memory
• Unknown - A Mamba-based foundation model for materials

Multimodality

Multimodal Integration in Chemistry

• 2024 - Unraveling Molecular Structure: A Multimodal Spectroscopic Dataset for Chemistry
• 2024 - Spectro: A multi-modal approach for molecule elucidation using IR and NMR data
• 2024 - Elucidating Structures from Spectra Using Multimodal Embeddings and Discrete Optimization
• 2024 - Seeing and Understanding: Bridging Vision with Chemical Knowledge Via ChemVLM
• 2024 - Probing the limitations of multimodal language models for chemistry and materials research
• 2023 - MolXPT: Wrapping Molecules with Text for Generative Pre-training
• 2023 - Multi-modal molecule structure–text model for text-based retrieval and editing
• 2023 - CLOOME: contrastive learning unlocks bioimaging databases for queries with chemical structures
• 2023 - InstructMol: Multi-Modal Integration for Building a Versatile and Reliable Molecular Assistant in Drug Discovery
• 2022 - Translation between Molecules and Natural Language
• 2022 - Galactica: A large language model for science

Optimizations

Mixture-of-Experts

• 2025 - UMA: A Family of Universal Models for Atoms
• 2025 - Multi-view Mixture-of-Experts for Predicting Molecular Properties Using SMILES, SELFIES, and Graph-Based Representations
• 2024 - SciDFM: A Large Language Model with Mixture-of-Experts for Science

Quantization and Mixed Precision

• 2017 - Mixed precision training

Parameter-Efficient Tuning

• 2022 - Lora: Low-rank adaptation of large language models

Distillation

• 2025 - Towards Fast, Specialized Machine Learning Force Fields: Distilling Foundation Models via Energy Hessians
• 2024 - Wisdom of Committee: Distilling from Foundation Model to Specialized Application Model
• 2019 - DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
• 2015 - Distilling the knowledge in a neural network

Model Level Adaptation

• 2025 - Integrating chemistry knowledge in large language models via prompt engineering
• 2025 - A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists
• 2023 - Palm: Scaling language modeling with pathways
• 2023 - GPT-4 Technical Report
• 2023 - ChatGPT Chemistry Assistant for Text Mining and Prediction of MOF Synthesis
• 2022 - Chain-of-thought prompting elicits reasoning in large language models
• 2022 - {LIFT: Language-Interfaced Fine-Tuning for Non-language Machine Learning Tasks}
• 2020 - Language models are few-shot learners
• 2019 - Language Models are Unsupervised Multitask Learners

System-level Integration: Agents

Core Components of an Agentic System

• 2025 - ORGANA: a robotic assistant for automated chemistry experimentation and characterization
• 2025 - Browsecomp: A simple yet challenging benchmark for browsing agents
• 2024 - Mle-bench: Evaluating machine learning agents on machine learning engineering
• 2024 - Augmenting large language models with chemistry tools
• 2024 - Language agents achieve superhuman synthesis of scientific knowledge
• 2023 - Toolformer: Language models can teach themselves to use tools
• 2023 - Autonomous chemical research with large language models
• 2023 - Chemist-X: large language model-empowered agent for reaction condition recommendation in chemical synthesis
• 2022 - Talm: Tool augmented language models
• 2020 - The next decade in AI: four steps towards robust artificial intelligence
• 2020 - Retrieval-augmented generation for knowledge-intensive nlp tasks

Approaches for building Agentic System

• 2025 - El Agente: An Autonomous Agent for Quantum Chemistry
• 2025 - Provence: efficient and robust context pruning for retrieval-augmented generation
• 2025 - How to Fix Your Context
• 2024 - ChatDev: Communicative Agents for Software Development
• 2024 - Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate
• 2024 - Can Long-Context Language Models Subsume Retrieval, RAG, SQL, and More?
• 2023 - React: Synergizing reasoning and acting in language models
• 2023 - Autogen: Enabling next-gen llm applications via multi-agent conversation
• 2023 - Improving factuality and reasoning in language models through multiagent debate
• 2023 - AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors
• 2020 - Emergent multi-agent communication in the deep learning era

Evaluations

The Evolution of Model Evaluation

• 2025 - A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists
• 2025 - Assessing the Chemical Intelligence of Large Language Models
• 2025 - MolLangBench: A Comprehensive Benchmark for Language-Prompted Molecular Structure Recognition, Editing, and Generation
• 2024 - Examining the robustness of LLM evaluation to the distributional assumptions of benchmarks
• 2024 - SciKnowEval: Evaluating Multi-level Scientific Knowledge of Large Language Models
• 2024 - Chemllm: A chemical large language model
• 2024 - LAB-Bench: Measuring Capabilities of Language Models for Biology Research
• 2024 - LabSafety Bench: Benchmarking LLMs on Safety Issues in Scientific Labs
• 2024 - LlaSMol: Advancing Large Language Models for Chemistry with a Large-Scale, Comprehensive, High-Quality Instruction Tuning Dataset
• 2024 - Probing the limitations of multimodal language models for chemistry and materials research
• 2024 - Can {LLMs Solve Molecule Puzzles? A Multimodal Benchmark for Molecular Structure Elucidation}
• 2024 - SciAssess: Benchmarking LLM Proficiency in Scientific Literature Analysis
• 2023 - Post Turing: Mapping the landscape of LLM Evaluation
• 2023 - MaScQA: A Question Answering Dataset for Investigating Materials Science Knowledge of Large Language Models
• 2023 - CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society
• 2023 - What can Large Language Models do in chemistry? A comprehensive benchmark on eight tasks
• 2018 - Model Evaluation, Model Selection, and Algorithm Selection in Machine Learning
• 2018 - MoleculeNet: a benchmark for molecular machine learning

Design of Evaluations

• 2025 - A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists
• 2024 - The dangers of using proprietary LLMs for research
• 2024 - A Survey of Useful LLM Evaluation
• 2023 - Post Turing: Mapping the landscape of LLM Evaluation
• 2022 - Why big data and compute are not necessarily the path to big materials science
• 2019 - Anthropogenic biases in chemical reaction data hinder exploratory inorganic synthesis

Evaluation Methodologies

• 2025 - Lessons from the trenches on evaluating machine-learning systems in materials science
• 2025 - A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists
• 2025 - Why Chatbots Keep Beating the Tests
• 2025 - NeurIPS - Open Polymer Prediction 2025
• 2024 - LAB-Bench: Measuring Capabilities of Language Models for Biology Research
• 2024 - Third Party Model Evaluations
• 2023 - CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society
• 2023 - SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
• 2023 - PromptRobust: Towards Evaluating the Robustness of Large Language Models on Adversarial Prompts
• 2023 - Certifying LLM Safety against Adversarial Prompting
• 2023 - MART: Improving LLM Safety with Multi-round Automatic Red-Teaming
• 2023 - GPT-4 Technical Report
• 2022 - Defining and Characterizing Reward Hacking
• 2022 - Red Teaming Language Models with Language Models
• 2022 - Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned
• 2022 - Dual use of artificial-intelligence-powered drug discovery
• 2020 - Deployment of Artificial Intelligence in Real-World Practice: Opportunity and Challenge
• 2018 - MoleculeNet: a benchmark for molecular machine learning
• 2005 - A decade of CASP: progress, bottlenecks and prognosis in protein structure prediction
• 2000 - A test of crystal structure prediction of small organic molecules

Future Directions

• 2025 - BixBench: a Comprehensive Benchmark for LLM-based Agents in Computational Biology
• 2025 - Lifting the benchmark iceberg with item-response theory
• 2024 - Agents for self-driving laboratories applied to quantum computing
• 2024 - Autonomous Microscopy Experiments through Large Language Model Agents
• 2024 - Augmenting large language models with chemistry tools
• 2023 - Autonomous chemical research with large language models
• 2023 - Generative Language Models and Automated Influence Operations: Emerging Threats and Potential Mitigations
• 2020 - Computer Vision for Recognition of Materials and Vessels in Chemistry Lab Settings and the Vector-LabPics Data Set
• 2005 - A decade of CASP: progress, bottlenecks and prognosis in protein structure prediction

Applications

Existing GPMs for Chemical Science

• 2025 - DARWIN 1.5: Large Language Models as Materials Science Adapted Learners
• 2025 - Training a Scientific Reasoning Model for Chemistry
• 2024 - nach0: Multimodal natural and chemical languages foundation model
• 2024 - From Tokens to Materials: Leveraging Language Models for Scientific Discovery
• 2024 - Foundational Large Language Models for Materials Research
• 2024 - ChemDFM: A Large Language Foundation Model for Chemistry
• 2024 - Chemllm: A chemical large language model
• 2023 - Unifying Molecular and Textual Representations via Multi-task Language Modelling
• 2020 - Predicting Chemical Properties using Self-Attention Multi-task Learning based on SMILES Representation

Knowledge Gathering

• 2025 - From text to insight: large language models for chemical data extraction
• 2024 - Language agents achieve superhuman synthesis of scientific knowledge
• 2022 - Quantifying the advantage of domain-specific pre-training on named entity recognition tasks in materials science
• 2021 - Slowed canonical progress in large fields of science
• 2021 - Automated Chemical Reaction Extraction from Scientific Literature
• 2019 - SciBERT: A Pretrained Language Model for Scientific Text
• 2018 - Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs
• 2017 - Enriching word vectors with subword information

Structured Data Extraction

• 2025 - From text to insight: large language models for chemical data extraction
• 2025 - LLM-as-Judge Meets LLM-as-Optimizer: Enhancing Organic Data Extraction Evaluations Through Dual LLM Approaches
• 2025 - Automated Data Extraction from Solar Cell Literature Using Large Language Models
• 2025 - MechBERT: Language Models for Extracting Chemical and Property Relationships about Mechanical Stress and Strain
• 2025 - A large language models-guided grand canonical DFT framework for accelerating the discovery of efficient electrocatalysts
• 2024 - Suitability of large language models for extraction of high-quality chemical reaction dataset from patent literature
• 2024 - Using machine-learning and large-language-model extracted data to predict copolymerizations
• 2024 - Data extraction from polymer literature using large language models
• 2024 - Extracting accurate materials data from research papers with conversational language models and prompt engineering
• 2024 - Reconstructing the materials tetrahedron: challenges in materials information extraction
• 2024 - An Autonomous Large Language Model Agent for Chemical Literature Data Mining
• 2024 - ChatMOF: an artificial intelligence system for predicting and generating metal-organic frameworks using large language models
• 2024 - Image and data mining in reticular chemistry powered by GPT-4V
• 2024 - Probing the limitations of multimodal language models for chemistry and materials research
• 2023 - Automatic extraction of FAIR data from publications using LLM
• 2023 - ChatGPT Chemistry Assistant for Text Mining and Prediction of MOF Synthesis
• 2023 - Improving factuality and reasoning in language models through multiagent debate
• 2022 - BatteryBERT: A pretrained language model for battery database enhancement
• 2021 - The Open Reaction Database

Question Answering

• 2023 - The Semantic Scholar Open Data Platform

Experiment Planning

• 2023 - On the Role of Large Language Models in Planning
• 2022 - A deep reinforcement learning based method for real-time path planning and dynamic obstacle avoidance

Conventional Planning

• 2025 - ASKCOS: an open source software suite for synthesis planning
• 2018 - Chematica: a story of computer code that started to think like a chemist
• 2017 - Towards" alphachem": Chemical synthesis planning with tree search and deep neural network policies
• 2014 - A short review of chemical reaction database systems, computer-aided synthesis design, reaction prediction and synthetic feasibility
• 1972 - Computer-assisted synthetic analysis for complex molecules. Methods and procedures for machine generation of synthetic intermediates

LLMs to Decompose Problems into Plans

• 2025 - A Survey of Test-Time Compute: From Intuitive Inference to Deliberate Reasoning
• 2024 - Chain of thoughtlessness? an analysis of cot in planning
• 2024 - Planning in natural language improves llm search for code generation
• 2024 - Meta-Designing Quantum Experiments with Language Models
• 2023 - Reasoning with language model is planning with world model
• 2023 - Llm+ p: Empowering large language models with optimal planning proficiency
• 2023 - Dynamic planning with a llm
• 2011 - Thinking, Fast and Slow

Pruning of Search Spaces

• 2025 - Synergizing rag and reasoning: A systematic review
• 2025 - ORGANA: a robotic assistant for automated chemistry experimentation and characterization
• 2012 - Action selection for MDPs: Anytime AO versus UCT

Evaluation

• 2024 - Augmenting large language models with chemistry tools
• 2024 - Lota-bench: Benchmarking language-oriented task planners for embodied agents
• 2023 - Llm-planner: Few-shot grounded planning for embodied agents with large language models
• 2023 - BioPlanner: automatic evaluation of LLMs on protocol planning in biology

Experiment Execution

• 2025 - Toward Automated Simulation Research Workflow through LLM Prompt Engineering Design
• 2025 - DynaMate: leveraging AI‑agents for customized research workflows
• 2025 - El Agente: An Autonomous Agent for Quantum Chemistry
• 2025 - MDCrow: Automating Molecular Dynamics Workflows with Large Language Models
• 2022 - Software update: the ORCA program system, version 5.0
• 2019 - Organic synthesis in a modular robotic system driven by a chemical programming language

Compiled Automation

• 2025 - Autonomous platform for solution processing of electronic polymers
• 2025 - Lowering the Entrance Hurdle for Lab Automation: An Artificial Intelligence‐Supported, Interactive Robotic Arm for Automated, Repeated Testing Procedures
• 2024 - Universal chemical programming language for robotic synthesis repeatability
• 2024 - Delocalized, asynchronous, closed-loop discovery of organic laser emitters
• 2024 - DSL‑Xpert: LLM‑driven Generic DSL Code Generation
• 2024 - ProtoCode: Leveraging large language models (LLMs) for automated generation of machine-readable PCR protocols from scientific publications
• 2024 - An integrated self-optimizing programmable chemical synthesis and reaction engine
• 2023 - PyLabRobot: An open-source, hardware-agnostic interface for liquid-handling robots and accessories
• 2023 - Self-Driving Laboratory for Polymer Electronics
• 2023 - Autoprotocol Specification
• 2023 - Artificial intelligence driven design of catalysts and materials for ring opening polymerization using a domain‑specific language
• 2023 - XDL 2.0 Standard Specification
• 2023 - LLMs can generate robotic scripts from goal-oriented instructions in biological laboratory automation
• 2023 - Large language models for chemistry robotics
• 2023 - Digitizing chemical discovery with a Bayesian explorer for interpreting reactivity data
• 2022 - ULSA: Unified language of synthesis actions for the representation of inorganic synthesis protocols
• 2021 - Chemputation and the standardization of chemical informatics
• 2021 - Digitizing Chemistry Using the Chemical Processing Unit: From Synthesis to Discovery
• 2020 - A universal system for digitization and automatic execution of the chemical synthesis literature
• 2019 - Organic synthesis in a modular robotic system driven by a chemical programming language
• 2010 - BioCoder: A programming language for standardizing and automating biology protocols

Interpreted Automation

• 2024 - Augmenting large language models with chemistry tools
• 2023 - Autonomous chemical research with large language models
• 2022 - Do as i can, not as i say: Grounding language in robotic affordances
• 2022 - Language Models as Zero-Shot Planners: Extracting Actionable Knowledge for Embodied Agents

Hybrid Approaches

• 2025 - ORGANA: a robotic assistant for automated chemistry experimentation and characterization
• 2023 - Large language models for chemistry robotics

Comparison and Outlook

• 2024 - Self-Driving Laboratories for Chemistry and Materials Science
• 2022 - Autonomous Chemical Experiments: Challenges and Perspectives on Establishing a Self‑Driving Lab

Data Analysis

• 2022 - Making the collective knowledge of chemistry open and machine actionable
• 1988 - JCAMP-DX: A Standard Form for Exchange of Infrared Spectra in Computer Readable Form

Prompting

• 2025 - Large Language Models as Spectrographic Assistants: Opportunities and Challenges in Laboratory Data Analysis
• 2024 - Large Language Model-Informed X-ray Photoelectron Spectroscopy Data Analysis
• 2024 - Probing the limitations of multimodal language models for chemistry and materials research

Agentic Systems

Current Limitations

• 2024 - Scicode: A research coding benchmark curated by scientists

Reporting

From Data to Explanation

• 2025 - Human interpretable structure-property relationships in chemistry using explainable machine learning and large language models
• 2022 - Model agnostic generation of counterfactual explanations for molecules

Writing Assistance

• 2024 - Using artificial intelligence in academic writing and research: An essential productivity tool
• 2024 - Figura11y: Ai assistance for writing scientific alt text
• 2023 - Medical image captioning via generative pretrained transformers
• 2023 - GPT-4 as an Effective Zero-Shot Evaluator for Scientific Figure Captions
• 2023 - The use of artificial intelligence to improve the scientific writing of non-native English speakers
• 2021 - SciCap: Generating captions for scientific figures

Vision

• 2025 - The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search

Accelerating Applications

• 2020 - Self-referencing embedded strings (SELFIES): A 100% robust molecular string representation
• 2016 - Understanding molecular representations in machine learning: The role of uniqueness and target similarity
• 1991 - The crystallographic information file (CIF): a new standard archive file for crystallography
• 1988 - SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules

Property Prediction

• 2025 - Semantic Device Graphs for Perovskite Solar Cell Design
• 2025 - Large language models for scientific discovery in molecular property prediction
• 2025 - A framework to evaluate machine learning crystal stability predictions
• 2024 - Leveraging large language models for predictive chemistry
• 2024 - {LLaMP: Large Language Model Made Powerful for High-fidelity Materials Knowledge Retrieval and Distillation}
• 2023 - LLM-Prop: Predicting Physical And Electronic Properties Of Crystalline Solids From Their Text Descriptions
• 2022 - {QMugs, quantum mechanical properties of drug-like molecules}
• 2021 - The {Tox21 10K Compound Library: Collaborative Chemistry Advancing Toxicology}
• 2021 - Prediction of {Blood-Brain Barrier Penetration (BBBP) Based on Molecular Descriptors of the Free-Form and In-Blood-Form Datasets}
• 2020 - {ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Prediction}
• 2018 - MoleculeNet: a benchmark for molecular machine learning
• 2017 - {DLS-100 Solubility Dataset}
• 2016 - The {SIDER database of drugs and side effects}
• 2014 - {FreeSolv: a database of experimental and calculated hydration free energies, with input files}
• 2012 - In silico screening of carbon-capture materials

Prompting

• 2024 - Cross-Modal Learning for Chemistry Property Prediction: Large Language Models Meet Graph Machine Learning
• 2023 - Group SELFIES: a robust fragment-based molecular string representation
• 2022 - Chemberta-2: Towards chemical foundation models

Fine-Tuning

• 2025 - Data-efficient fine-tuning of foundational models for first-principles quality sublimation enthalpies
• 2024 - On the scalability of gnns for molecular graphs
• 2023 - From Molecules to Materials: Pre-training Large Generalizable Models for Atomic Property Prediction
• 2022 - MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields
• 2020 - Exploring the limits of transfer learning with a unified text-to-text transformer
• 2019 - Robocrystallographer: automated crystal structure text descriptions and analysis
• Unknown - Exploring BERT for Reaction Yield Prediction: Evaluating the Impact of Tokenization, Molecular Representation, and Pretraining Data Augmentation
• Unknown - A Mamba-based foundation model for materials

Agents

• 2023 - 14 examples of how LLMs can transform materials science and chemistry: a reflection on a large language model hackathon

Core Limitations

• 2023 - Neural scaling of deep chemical models
• 2021 - Compositionally restricted attention-based network for materials property predictions
• Unknown - The Promises and Pitfalls of Language Models for Structured Numerical Data

Molecular and Material Generation

• 2024 - A novel molecule generative model of VAE combined with Transformer for unseen structure generation

Generation

• 2025 - Large Language Models are in-Context Molecule Learners
• 2025 - Property Enhanced Instruction Tuning for Multi-task Molecule Generation with Large Language Models
• 2025 - Aligning Transformers with Continuous Feedback via Energy Rank Alignment
• 2025 - CrystalFormer-RL: Reinforcement Fine-Tuning for Materials Design
• 2025 - RAG-Enhanced Collaborative LLM Agents for Drug Discovery
• 2024 - Crossing New Frontiers: Knowledge-Augmented Large Language Model Prompting for Zero-Shot Text-Based De Novo Molecule Design
• 2024 - MolReFlect: Towards In-Context Fine-grained Alignments between Molecules and Texts
• 2024 - 3M-Diffusion: Latent Multi-Modal Diffusion for Language-Guided Molecular Structure Generation
• 2024 - FlowLLM: Flow Matching for Material Generation with Large Language Models as Base Distributions
• 2024 - Multimodal Large Language Models for Inverse Molecular Design with Retrosynthetic Planning
• 2024 - FINE-TUNING POCKET-CONDITIONED 3D MOLECULE GENERATION VIA REINFORCEMENT LEARNING
• 2024 - Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization
• 2024 - Space Group Informed Transformer for Crystalline Materials Generation
• 2024 - dZiner: Rational Inverse Design of Materials with AI Agents
• 2023 - What can Large Language Models do in chemistry? A comprehensive benchmark on eight tasks
• 2023 - MolCA: Molecular Graph-Language Modeling with Cross-Modal Projector and Uni-Modal Adapter
• 2023 - Rewarded soups: towards Pareto-optimal alignment by interpolating weights fine-tuned on diverse rewards
• 2022 - Translation between Molecules and Natural Language
• 2021 - Text2Mol: Cross-Modal Molecule Retrieval with Natural Language Queries

Validation

• 2025 - A framework to evaluate machine learning crystal stability predictions
• 2022 - A flexible and scalable scheme for mixing computed formation energies from different levels of theory
• 2019 - Beware of plausible predictions of fantasy materials
• 2019 - GuacaMol: Benchmarking Models for de Novo Molecular Design
• 2018 - Fréchet ChemNet Distance: A Metric for Generative Models for Molecules in Drug Discovery

Retrosynthesis

• 2025 - Explainable Synthesizability Prediction of Inorganic Crystal Polymorphs Using Large Language Models

Accelerating Applications

Automating the Scientific Workflow

• 2025 - Agent Laboratory: Using LLM Agents as Research Assistants

Coding and ML Applications of AI Scientists

• 2025 - Towards an AI co-scientist
• 2025 - The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search
• 2025 - AgentRxiv: Towards Collaborative Autonomous Research
• 2025 - PaperBench: Evaluating AI's Ability to Replicate AI Research
• 2025 - Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents
• 2025 - AlphaEvolve: A coding agent for scientific and algorithmic discovery
• 2024 - Self-Reflection in LLM Agents: Effects on Problem-Solving Performance
• 2024 - Mle-bench: Evaluating machine learning agents on machine learning engineering
• 2024 - Automated Design of Agentic Systems
• 2023 - MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation

Chemistry and Related Fields

• 2025 - Robin: A multi-agent system for automating scientific discovery
• 2025 - El Agente: An Autonomous Agent for Quantum Chemistry
• 2025 - A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists
• 2025 - Training a Scientific Reasoning Model for Chemistry
• 2024 - The perpetual motion machine of AI-generated data and the distraction of ChatGPT as a ‘scientist’

Are these Systems Capable of Real Autonomous Research?

• 2025 - Zochi Publishes A* Paper
• 2025 - When AI Co-Scientists Fail: SPOT-a Benchmark for Automated Verification of Scientific Research

Implications of GPMs: Education, Safety, and Ethics

Education

Vision

• 2025 - The role of large language models in personalized learning: a systematic review of educational impact
• 2025 - The effect of ChatGPT on students' learning performance, learning perception, and higher-order thinking: insights from a meta-analysis
• 2024 - AI Agents and Education: Simulated Practice at Scale
• 2024 - Instructors as Innovators: A future-focused approach to new AI learning opportunities, with prompts
• 2024 - Large Language Models are Catalyzing Chemistry Education
• 2024 - Grading assistance for a handwritten thermodynamics exam using artificial intelligence: An exploratory study
• 2024 - Towards Scalable Automated Grading: Leveraging Large Language Models for Conceptual Question Evaluation in Engineering

Current Status

• 2025 - Large Language Models (LLMs) as Graphing Tools for Advanced Chemistry Education and Research
• 2025 - Unlocking Scientific Concepts: How Effective Are LLM-Generated Analogies for Student Understanding and Classroom Practice?
• 2025 - TutorLLM: Customizing Learning Recommendations with Knowledge Tracing and Retrieval-Augmented Generation
• 2025 - LearnMate: Enhancing Online Education with LLM-Powered Personalized Learning Plans and Support
• 2025 - Claude for Education | Partnering with Universities on Responsible AI
• 2023 - Exploring the use of large language models (LLMs) in chemical engineering education: Building core course problem models with Chat-GPT
• 2023 - 14 examples of how LLMs can transform materials science and chemistry: a reflection on a large language model hackathon

Outlook and Limitations

• 2025 - Will the Humanities Survive Artificial Intelligence?
• 2025 - Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task
• 2025 - Learning Alone: Language Models, Overreliance, and the Goals of Education
• 2025 - The role of large language models in personalized learning: a systematic review of educational impact
• 2025 - We Have to Really Rethink the Purpose of Education

Safety

• 2024 - The Reality of AI and Biorisk
• 2023 - Levels of AGI for Operationalizing Progress on the Path to AGI
• 2009 - The tacit dimension
• Unknown - Statement on AI Risk CAIS

Evaluating Risk Amplification in the Chemical Discovery Cycle

• 2025 - System Card: Claude Opus 4 & Claude Sonnet 4
• 2025 - The Risks and Rewards of Embodying Artificial Intelligence with Cloud-Based Laboratories
• 2025 - Security Forecast -- AI 2027
• 2024 - The Reality of AI and Biorisk
• 2024 - Building an early warning system for LLM-aided biological threat creation
• 2024 - Challenges in Guardrailing Large Language Models for Science
• 2023 - Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools
• 2023 - Survey of Hallucination in Natural Language Generation

Existing Approaches to Safety

• 2025 - Open Problems in Machine Unlearning for AI Safety
• 2025 - OS-Harm: A Benchmark for Measuring Safety of Computer Use Agents
• 2025 - Agentic Misalignment: How LLMs Could be an Insider Threat
• 2024 - Augmenting large language models with chemistry tools
• 2024 - Prioritizing Safeguarding Over Autonomy: Risks of LLM Agents for Science
• 2023 - Sparse Autoencoders Find Highly Interpretable Features in Language Models
• 2023 - Is This the Subspace You Are Looking for? An Interpretability Illusion for Subspace Activation Patching
• 2023 - Autonomous chemical research with large language models
• 2022 - Constitutional AI: Harmlessness from AI Feedback
• Unknown - Stealing User Prompts from Mixture of Experts

Solutions

• 2025 - International AI Safety Report
• 2024 - AI and biosecurity: The need for governance
• 2024 - Safety Considerations for Chemical and/or Biological AI Models
• 2024 - Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (Text with EEA relevance)
• 2024 - CERN publishes its first Nuclear Safeguards Policy
• 2023 - International Governance of Civilian AI: A Jurisdictional Certification Approach
• 2016 - Unethical Research: How to Create a Malevolent Artificial Intelligence
• Unknown - Career Update: Google DeepMind -> Anthropic

Ethics

• Unknown - The Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence

Environmental Impact and Climate Ethics

• 2025 - Considering the Ethics of Large Machine Learning Models in the Chemical Sciences
• 2025 - Google undercounts its carbon emissions, report finds
• 2024 - The Obscene Energy Demands of A.I.
• 2021 - Data centre water consumption
• 2019 - Energy and Policy Considerations for Deep Learning in NLP
• Unknown - The carbon footprint of computational research

Copyright Infringement and Plagiarism Concerns

• 2024 - Intellectual property rights at the training, development and generation stages of Large Language Models
• 2023 - Written Evidence to [Committee Name]
• 2023 - Artificial intelligence in scientific writing: a friend or a foe?
• 2023 - The dangers of using large language models for peer review
• 2021 - On the dangers of stochastic parrots: Can language models be too big?

Bias and Discrimination

• 2025 - Considering the Ethics of Large Machine Learning Models in the Chemical Sciences
• 2024 - Demographic Bias of Expert-Level Vision-Language Foundation Models in Medical Imaging
• 2023 - Large language models propagate race-based medicine
• 2023 - Algorithmic fairness in artificial intelligence for medicine and healthcare
• 2023 - Bias in AI-based models for medical applications: challenges and mitigation strategies
• 2019 - Value-laden Disciplinary Shifts in Machine Learning
• 2019 - Invisible women: exposing data bias in a world designed for men

Democratization of Power

• 2025 - The A.I. Race Is Splitting the World Into Haves and Have-Nots

Outlook and Conclusions

• 2025 - Exclusive: Start-up FutureHouse debuts powerful AI ‘reasoning model’ for science
• 2014 - Machine learning: The high interest credit card of technical debt

optimizers

LLMs as Optimizers

• 2025 - Adaptive representation of molecules and materials in Bayesian optimization
• 2025 - Language-Based Bayesian Optimization Research Assistant (BORA)
• 2024 - Sequential closed-loop Bayesian optimization as a guide for organic molecular metallophotocatalyst formulation discovery
• 2024 - InstructZero: Efficient Instruction Optimization for Black-Box Large Language Models
• 2023 - A Brief Introduction to Chemical Reaction Optimization
• 2023 - Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution
• 2023 - Large Language Models as Optimizers
• 2023 - BoChemian: Large Language Model Embeddings for Bayesian Optimization of Chemical Reactions
• 2021 - Gryffin: An algorithm for Bayesian optimization of categorical variables informed by expert knowledge
• 2021 - Bayesian reaction optimization as a tool for chemical synthesis
• 2020 - Constrained Bayesian optimization for automatic chemical design using variational autoencoders

LLMs as Surrogate Models

• 2022 - Sample Efficiency Matters: A Benchmark for Practical Molecular Optimization

LLMs as Next Candidate Generators

• 2025 - Generative Design of Functional Metal Complexes Utilizing the Internal Knowledge and Reasoning Capability of Large Language Models
• 2025 - Towards Fast, Specialized Machine Learning Force Fields: Distilling Foundation Models via Energy Hessians

LLMs as Prior Knowledge Sources

How to Face Optimization Problems?

• 2025 - GOLLuM: Gaussian Process Optimized LLMs - Reframing LLM Finetuning through Bayesian Optimization
• 2025 - LLM-Augmented Chemical Synthesis and Design Decision Programs
• 2025 - Efficient Evolutionary Search Over Chemical Space with Large Language Models
• 2024 - A Sober Look at LLMs for Material Discovery: Are They Actually Good for Bayesian Optimization Over Molecules?
• 2023 - Bayesian Optimization of Catalysis With In-Context Learning

hypothesis

Hypothesis Generation

• 2025 - Sparks of Science: Hypothesis Generation Using Structured Paper Data
• 2022 - The Roles of a Secondary Data Analytics Tool and Experience in Scientific Hypothesis Generation in Clinical Research: Protocol for a Mixed Methods Study
• 2019 - AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence
• 2018 - A hypothesis can't be right unless it can be proven wrong
• 2017 - What Goes Up... Gravity and Scientific Method
• 2017 - Open-endedness: The Last Grand Challenge You've Never Heard Of
• 2015 - Why Greatness Cannot Be Planned: The Myth of the Objective
• 1999 - The Principia: Mathematical Principles of Natural Philosophy
• 1959 - The Logic of Scientific Discovery

Initial Sparks

• 2025 - Forecasting high-impact research topics via machine learning on evolving knowledge graphs
• 2025 - CodeScientist: End-to-End Semi-Automated Scientific Discovery with Code-based Experimentation
• 2025 - Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agents
• 2025 - DORA AI Scientist: Multi-agent Virtual Research Team for Scientific Exploration Discovery and Automated Report Generation
• 2025 - Robin: A multi-agent system for automating scientific discovery
• 2024 - Meta-Designing Quantum Experiments with Language Models
• 2024 - Interesting Scientific Idea Generation using Knowledge Graphs and LLMs: Evaluations with 100 Research Group Leaders
• 2024 - OMNI: Open-endedness via Models of human Notions of Interestingness

Chemistry-Focused Hypotheses

• 2024 - Are LLMs Ready for Real-World Materials Discovery?

Are LLMs Actually Capable of Novel Hypothesis Generation?

• 2025 - EXP-Bench: Can AI Conduct AI Research Experiments?
• 2025 - Tempest: Autonomous Multi-Turn Jailbreaking of Large Language Models with Tree Search
• 2025 - MOOSE-Chem2: Exploring LLM Limits in Fine-Grained Scientific Hypothesis Discovery via Hierarchical Search
• 2025 - The Ideation-Execution Gap: Execution Outcomes of LLM-Generated versus Human Research Ideas
• 2025 - AlphaEvolve: A coding agent for scientific and algorithmic discovery
• 2024 - Interesting Scientific Idea Generation using Knowledge Graphs and LLMs: Evaluations with 100 Research Group Leaders
• 2024 - Hypothesis-Generating Research
• 2020 - Autonomous discovery in the chemical sciences part I: Progress
• 2015 - Why Greatness Cannot Be Planned: The Myth of the Objective
• 2006 - The Millennium Prize Problems
• 1970 - Falsification and the Methodology of Scientific Research Programmes
• 1964 - Penicillin
• 1962 - The Structure of Scientific Revolutions
• 1959 - The Logic of Scientific Discovery
• 1929 - On the Antibacterial Action of Cultures of a Penicillium, with Special Reference to Their Use in the Isolation of B. influenzae