Attention Is All You Need
The Transformer authors
Transformers made modern large language model workflows possible.
Read the public workSpecial Thanks
AI tools are built on shared work.
AIPM exists because researchers, product teams, open communities, and conferences made AI useful for everyday project work. This page says thanks. It does not claim endorsement.
Foundational deep learning research
Geoffrey Hinton, Yoshua Bengio, and Yann LeCun
Their work helped build the neural network foundation behind today's AI systems.
Read the public workImageNet and data-centric benchmark culture
Fei-Fei Li and the ImageNet contributors
ImageNet showed how shared datasets and benchmarks can move a field forward.
Read the public workAlphaFold
Demis Hassabis, John Jumper, and the AlphaFold team
AlphaFold showed how AI can help with real scientific discovery.
Read the public workReinforcement learning foundations
Richard Sutton and Andrew Barto
Their reinforcement learning work shaped how agents learn from actions and rewards.
Read the public workGPT models, ChatGPT, APIs, and developer tooling
OpenAI research and product teams
Their public products made powerful AI available to many builders and teams.
Read the public workClaude, Constitutional AI, and safety-focused assistant design
Anthropic research and product teams
Their work helped popularize safer assistant behavior and practical prompt guidance.
Read the public workAI education for practitioners
Andrew Ng and DeepLearning.AI
Their courses and writing made machine learning and AI easier for working developers to learn.
Read the public workGPU computing and AI systems infrastructure
NVIDIA and the accelerated computing community
Modern AI depends on the hardware and systems that make large-scale training and inference practical.
Read the public workMeasuring AI progress and impact
Stanford HAI and the AI Index team
The AI Index gives builders, policymakers, and researchers data about how AI is changing.
Read the public workRecent conference signals
AI changes through papers, talks, workshops, datasets, replication, and products. AIPM tracks these communities so reusable skills stay close to real practice. Updated June 3, 2026.
NeurIPS 2025 invited speakersKyunghyun Cho's GRU and neural machine translation work, Yejin Choi's commonsense reasoning research, Melanie Mitchell's abstraction and analogy work, Andrew Saxe's theory of learning, Richard Sutton's reinforcement learning foundations, and Zeynep Tufekci's technology-and-society analysis.Reusable AI skills should be tested on hard cases, not only happy-path demos.NeurIPS announcementICLR 2026 keynotesMaja Mataric on human-centered AI and robotics, Max Welling on physics-to-AI-to-materials, Percy Liang on Marin and open frontier AI, Katie Bouman on imaging hidden science, Karen Adolph on infant learning, and Pablo Arbelaez on AI for open science.AIPM should make AI work easy to inspect and grounded in real user behavior.ICLR keynote announcementICML 2026 invited talksPascale Fung on conversational and ethical AI, Susan Athey on causal inference and AI economics, Sham Kakade on RL and deep learning theory, Aviv Regev on AI for biology, Verena Rieser on alignment and evaluation, and Arvind Narayanan on AI's social impact.Publishing should include evaluation, governance, and context before a skill is widely reused.ICML invited-talk announcementICLR 2026 outstanding papersTransformers are Inherently Succinct, LLMs Get Lost In Multi-Turn Conversation, and The Polar Express: Optimal Matrix Sign Methods and their Application to the Muon Algorithm.AIPM skills need multi-turn tests, clear versions, and honest notes about failure cases.ICLR outstanding-papers announcementCVPR 2025 keynotesHarry Shum on low-altitude airspace infrastructure, Laurens van der Maaten on the Llama herd of models, and Carolina Parada on Gemini Robotics and embodied AI.Reusable AI tooling should support more than simple prompt files as the registry grows.CVPR keynote announcement
How we keep this page fair
- Use public sources and direct links to papers, projects, or official pages.
- Separate appreciation from endorsement or partnership claims.
- Update names as new public work becomes important to AI builders.
- Credit communities, maintainers, and educators, not only famous founders.