This timeline traces the foundational milestones of Artificial Intelligence and Large Language Models, from early theoretical work to the emergence of modern LLMs. It covers key breakthroughs in neural networks, natural language processing, and the scaling of transformer models that define Epoch 1.
Alan Turing publishes 'Computing Machinery and Intelligence,' introducing the Turing Test as a criterion for machine intelligence. This seminal paper lays the philosophical groundwork for AI. #AI #philosophy #computing
John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon organize the Dartmouth Summer Research Project, officially launching AI as a field. The term 'Artificial Intelligence' is first used. #AI #history #computing
Frank Rosenblatt invents the Perceptron, an early neural network capable of simple pattern recognition. This marks the first implementation of a trainable artificial neuron. #AI #neuralnetworks #machinelearning
Joseph Weizenbaum at MIT creates ELIZA, a natural language processing program that simulates a psychotherapist. It demonstrates early conversational AI using pattern matching. #AI #chatbot #NLP
Terry Winograd develops SHRDLU, a program that understands and manipulates a virtual block world using natural language commands. It showcases early AI reasoning and language understanding. #AI #NLP #robotics
After reduced funding and unmet expectations, the first AI winter sets in. Research in neural networks and symbolic AI slows significantly due to limited computational power and funding cuts. #AI #history #setback
David Rumelhart, Geoffrey Hinton, and Ronald Williams publish a key paper on backpropagation, enabling multi-layer neural networks to learn. This reignites interest in neural networks. #AI #neuralnetworks #deeplearning
IBM's Deep Blue becomes the first computer to beat a reigning world chess champion, Garry Kasparov, in a match. This milestone demonstrates AI's potential in complex strategic games. #AI #chess #gaming
Geoffrey Hinton publishes a paper on deep belief networks and greedy layer-wise pretraining, sparking the deep learning revolution. This enables training deeper neural networks effectively. #AI #deeplearning #neuralnetworks
IBM's Watson defeats human champions Ken Jennings and Brad Rutter on the quiz show Jeopardy!. It showcases advanced natural language processing and question-answering capabilities. #AI #NLP #quiz
Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton develop AlexNet, a deep convolutional neural network that dramatically improves image classification. This marks a turning point for deep learning in computer vision. #AI #deeplearning #computervision
Tomas Mikolov and team at Google release Word2Vec, a neural network model that learns word embeddings capturing semantic relationships. It revolutionizes natural language processing by enabling vector representations of words. #AI #NLP #embeddings
Ian Goodfellow introduces Generative Adversarial Networks, where two neural networks compete to generate realistic data. GANs become a cornerstone for generative AI and image synthesis. #AI #GANs #generative
Bahdanau, Cho, and Bengio propose the attention mechanism for sequence-to-sequence models, improving machine translation. This innovation becomes fundamental for later transformer architectures. #AI #NLP #translation
DeepMind's AlphaGo becomes the first computer program to beat a professional Go player, Fan Hui, without handicaps. This milestone demonstrates AI mastery in a complex board game. #AI #gaming #deeplearning
AlphaGo defeats world champion Lee Sedol 4-1 in a five-game match, a landmark event for AI. The match is watched by millions and highlights the power of deep reinforcement learning. #AI #gaming #reinforcementlearning
Vaswani et al. from Google publish 'Attention Is All You Need,' introducing the Transformer model. This architecture, based solely on attention mechanisms, becomes the foundation for modern LLMs. #AI #transformers #NLP
Google releases BERT (Bidirectional Encoder Representations from Transformers), a pre-trained language model that achieves state-of-the-art results on 11 NLP tasks. BERT popularizes the 'pre-train then fine-tune' paradigm. #AI #NLP #pretraining
OpenAI releases the first Generative Pre-trained Transformer (GPT-1), a 117M parameter model. It demonstrates the effectiveness of unsupervised pre-training for language understanding and generation. #AI #LLM #generative
OpenAI announces GPT-2 with 1.5B parameters, but initially withholds the full model due to concerns about misuse. It generates coherent and contextually relevant text, sparking debates on AI safety. #AI #LLM #safety
OpenAI releases the full 1.5B parameter GPT-2 model after a staged release process. It becomes widely used for text generation and demonstrates the scaling benefits of larger language models. #AI #LLM #openairelease
OpenAI announces GPT-3 with 175B parameters, a massive leap in scale. It exhibits few-shot and zero-shot learning capabilities, generating human-like text across diverse tasks without fine-tuning. #AI #LLM #scaling
OpenAI launches a commercial API for GPT-3, allowing developers to integrate powerful language generation into applications. This marks the beginning of widespread LLM-based services. #AI #LLM #API
OpenAI announces DALL-E, a model that generates images from text descriptions using a 12B parameter transformer. It demonstrates multimodal AI capabilities combining language and vision. #AI #multimodal #generative
GitHub and OpenAI launch Copilot, an AI pair programmer powered by a version of GPT-3 fine-tuned on code. It assists developers by suggesting code snippets and functions. #AI #coding #productivity
Google announces PaLM (Pathways Language Model) with 540B parameters, achieving strong performance on reasoning and language tasks. It highlights the benefits of scaling and efficient training. #AI #LLM #google
OpenAI releases ChatGPT, a conversational agent based on GPT-3.5 fine-tuned with reinforcement learning from human feedback. It quickly gains over 100 million users, becoming a global phenomenon. #AI #chatbot #LLM
Microsoft announces a multi-billion dollar investment in OpenAI, deepening the partnership to integrate AI into products like Azure, Office, and Bing. This signals major corporate commitment to LLMs. #AI #investment #partnership
Microsoft launches the new Bing with an integrated ChatGPT-like chatbot, later named Copilot. It combines search and conversational AI, marking a shift in how users interact with search engines. #AI #search #chatbot
OpenAI releases GPT-4, a multimodal model capable of processing text and images. It demonstrates improved reasoning, creativity, and safety, setting new benchmarks for LLM performance. #AI #LLM #multimodal
Google launches Bard, a conversational AI chatbot initially based on LaMDA, later upgraded to the Gemini model. It competes with ChatGPT and integrates with Google's ecosystem. #AI #chatbot #google
Meta releases Llama 2, an open-source large language model with up to 70B parameters. It is made available for research and commercial use, spurring innovation in the open-source AI community. #AI #opensource #LLM
At its first developer conference, OpenAI announces customizable GPTs and GPT-4 Turbo with a 128K context window. This expands the capabilities and accessibility of LLMs for developers. #AI #LLM #developer
OpenAI announces Sora, a text-to-video model capable of generating realistic and imaginative scenes. It represents a major advancement in generative AI beyond text and images. #AI #video #generative
Anthropic releases Claude 3, a family of LLMs including Haiku, Sonnet, and Opus. It emphasizes safety and performance, achieving state-of-the-art results on various benchmarks. #AI #LLM #safety
OpenAI announces GPT-4o ('omni'), a multimodal model that natively integrates text, vision, and audio. It enables real-time voice conversations and improved vision understanding. #AI #multimodal #voice
Meta releases Llama 3.1 with up to 405B parameters, the largest open-source LLM at the time. It matches or exceeds proprietary models on many benchmarks, democratizing access to advanced AI. #AI #opensource #LLM
OpenAI releases the o1 model, which uses chain-of-thought reasoning to improve performance on complex math, science, and coding tasks. It represents a new paradigm in LLM reasoning. #AI #reasoning #LLM
The European Union passes the AI Act, the world's first comprehensive AI regulation. It categorizes AI systems by risk and imposes requirements for transparency and safety, impacting LLM deployment. #AI #regulation #EU
Google announces Gemini 2.0, a next-generation multimodal AI model with enhanced capabilities in reasoning, coding, and agentic tasks. It powers new features across Google products. #AI #multimodal #google
Chinese AI lab DeepSeek releases DeepSeek-R1, an open-source reasoning model that rivals OpenAI's o1. It demonstrates strong performance in math and coding, highlighting global competition in LLMs. #AI #opensource #reasoning
Multiple companies launch AI agent platforms that combine LLMs with tool use and autonomous task execution. This marks a shift from conversational AI to proactive, goal-oriented agents. #AI #agents #automation
Anthropic releases Claude 4, featuring improved safety, longer context windows, and better reasoning. It sets new standards for responsible AI development and deployment. #AI #LLM #safety
OpenAI announces GPT-5, a major upgrade with enhanced multimodal capabilities, improved reasoning, and larger context. It is expected to push the boundaries of LLM performance further. #AI #LLM #multimodal
World leaders and AI researchers convene for a global summit on AI safety, focusing on risks from advanced LLMs. Agreements are made on voluntary safety standards and transparency measures. #AI #safety #governance
Several open-source LLMs, including Llama 4 and Mistral Large, achieve performance comparable to proprietary models like GPT-4. This democratization accelerates AI research and application. #AI #opensource #LLM
LLM-powered personalized tutoring systems are adopted by major school districts, providing adaptive learning experiences. This marks a significant real-world impact of AI in education. #AI #education #tutoring
A novel co-written by an LLM and a human author wins a major literary award, sparking debate about creativity and authorship. This event highlights the evolving role of AI in creative fields. #AI #literature #creativity
Several LLMs achieve context windows of 10 million tokens, enabling processing of entire books or large codebases. This breakthrough expands applications in document analysis and long-form reasoning. #AI #LLM #context
Over 50 countries sign a global treaty on AI governance, establishing binding rules for high-risk AI systems. This marks a milestone in international cooperation on AI safety and ethics. #AI #regulation #global