The CHIPS and Science Act funds US chip manufacturing
Signed in August 2022, the CHIPS and Science Act appropriated roughly $50 billion to bring advanced semiconductor manufacturing back to the United States.
Stability AI releases Stable Diffusion to the public
Stability AI publicly releases Stable Diffusion, an open text-to-image model with downloadable weights and code.
Character.AI opens, and millions start talking to AI personas
Character.AI opened its beta in September 2022; two months later it generated a billion words a day across 350,000 user-made characters.
OpenAI releases Whisper speech recognition
OpenAI releases Whisper, an open multilingual speech recognition model trained on large-scale weakly supervised audio.
EU Proposes the AI Liability Directive (2022)
The European Commission proposes rules easing the burden of proof for people seeking compensation for harm caused by AI systems.
LangChain frames the toolkit for building LLM applications
LangChain, started by Harrison Chase in late 2022, became the most widely used open-source framework for chaining LLMs with tools and data.
White House Blueprint for an AI Bill of Rights
The US White House released five non-binding principles meant to protect the public from harms of automated systems.
AlphaTensor discovers faster matrix multiplication algorithms
DeepMind's AlphaTensor used reinforcement learning to find new matrix multiplication algorithms, beating Strassen's 1969 method.
US Advanced-Computing Chip Export Controls (2022)
On 7 October 2022 the US Bureau of Industry and Security imposed sweeping export controls on advanced AI chips and chipmaking equipment bound for China.
ANA Avatar XPRIZE awarded
Team NimbRo won the $5M ANA Avatar XPRIZE for a telepresence robot that let an operator see, hear, and act in a distant location.
LlamaIndex (GPT Index) gives LLMs a way to read your data
In November 2022 Jerry Liu released GPT Index, later LlamaIndex, an open-source framework for connecting language models to private data via retrieval.
Refik Anadol's Unsupervised brings AI art to MoMA's lobby
MoMA installed Refik Anadol's Unsupervised, a machine-learning work that endlessly re-imagines the museum's collection of 138,151 artworks.
Cicero plays Diplomacy at a human level
Meta's Cicero (2022) was the first AI to reach human-level play in Diplomacy, a game of negotiation and trust, pairing a language model with strategic planning.
China Issues Deep Synthesis (Deepfake) Rules (2022)
China issues provisions requiring labeling and consent for AI-generated synthetic media, in force from January 2023, among the world's first deepfake rules.
OpenAI launches ChatGPT
OpenAI releases ChatGPT, a conversational AI that reaches mass adoption and ignites the modern generative-AI boom.
Riffusion makes music by fine-tuning Stable Diffusion on spectrograms
Riffusion repurposed the Stable Diffusion image model to paint spectrograms, then turned those pictures of sound into music.
A Princeton student launches GPTZero to detect AI writing
GPTZero, an AI-writing detector built over winter break by Princeton senior Edward Tian, went viral on January 1, 2023.
ICML bans LLM-generated paper text in its 2023 policy
ICML 2023 barred papers containing LLM-generated text unless it was part of the experimental analysis, while still permitting light editing.
New York City schools ban ChatGPT, then reverse course
NYC public schools blocked ChatGPT in January 2023, then lifted the ban four months later, calling the initial fear knee-jerk.
DoD updates Directive 3000.09 on autonomy in weapons
In January 2023 the Pentagon reissued its policy on autonomy in weapon systems, requiring appropriate human judgment over the use of force.
NIST releases the AI Risk Management Framework (2023)
NIST published its voluntary AI Risk Management Framework, organized around four functions: Govern, Map, Measure, and Manage.
Italy's Garante bans Replika from processing Italian users' data
In February 2023 Italy's data regulator ordered Replika to stop processing Italian users' data, citing missing age checks and risks to children.
The first REAIM summit on responsible military AI
In February 2023 the Netherlands and South Korea convened the first global summit on responsible AI in the military domain, ending in a joint call to action.
The Political Declaration on Responsible Military Use of AI
A US-led, non-binding declaration setting out norms for responsible military use of AI and autonomy, launched at the 2023 REAIM summit.
Copyright Office limits registration of an AI-illustrated comic
In February 2023 the US Copyright Office ruled that Midjourney images in the comic Zarya of the Dawn were not protected by copyright.
Notion AI becomes generally available to all users
Notion opened its AI writing assistant to everyone in February 2023 after a ten-week alpha that drew nearly two million waitlist signups.
Meta releases LLaMA
Meta releases LLaMA, a family of foundational large language models, accelerating the open-weights movement.
Snapchat puts a GPT chatbot in front of hundreds of millions with My AI
In February 2023 Snapchat launched My AI, a GPT-powered chatbot pinned in the app, one of the first companion bots inside a mass social network.
Grammarly adds generative AI with GrammarlyGO
Grammarly announced GrammarlyGO in March 2023, extending its editing tool into generative drafting, rewriting, and AI-composed email replies.
llama.cpp puts large language models on ordinary laptops
Georgi Gerganov's llama.cpp, released in March 2023, ran LLaMA in plain C/C++ with quantization, letting large models run on consumer laptops without a GPU.
Anthropic introduces Claude
Anthropic launches Claude, a next-generation AI assistant built around helpful, honest, and harmless training.
Be My Eyes builds a GPT-4 visual assistant for blind users
Be My Eyes launched Virtual Volunteer in 2023, the first consumer use of GPT-4's image understanding, describing scenes for blind and low-vision users.
Casetext launches CoCounsel, the first GPT-4 legal assistant
Casetext launched CoCounsel in March 2023, an AI legal assistant built on early access to GPT-4 to do research, document review, and memo drafting.
Duolingo launches Duolingo Max, a GPT-4 subscription tier
Duolingo introduced a premium tier powered by GPT-4 in March 2023, adding conversation roleplay and answer explanations to language lessons.
Khan Academy launches Khanmigo, a GPT-4 tutor
Khan Academy unveiled Khanmigo in March 2023, a GPT-4 tutor that guides students with questions instead of handing over answers.
Khan Academy launches Khanmigo, a GPT-4 tutor
On March 14, 2023, Khan Academy unveiled Khanmigo, a GPT-4-powered Socratic tutor and teaching assistant piloted in classrooms.
OpenAI releases GPT-4
OpenAI releases GPT-4, a multimodal model scoring in the top tiers of professional exams and marking a major capability jump.
Zhipu AI and Tsinghua open-source ChatGLM-6B
On March 14, 2023, Zhipu AI and Tsinghua released ChatGLM-6B, a bilingual chat model that ran on a single consumer GPU via INT4 quantization.
Microsoft announces Microsoft 365 Copilot for Office apps
Microsoft unveiled Copilot for Word, Excel, PowerPoint, Outlook, and Teams in March 2023, grounding a GPT model in a company's own data via Microsoft Graph.
Nuance DAX Express brings GPT-4 ambient documentation to the clinic
In March 2023 Nuance and Microsoft unveiled DAX Express, billed as the first clinical-documentation app to combine ambient AI with OpenAI's GPT-4.
Adobe launches Firefly, generative AI trained on licensed content
In March 2023 Adobe unveiled Firefly, a generative image AI it positioned as commercially safe by training only on licensed and public-domain content.
The Pause Giant AI Experiments Letter
A March 2023 open letter from the Future of Life Institute called on all AI labs to pause training systems more powerful than GPT-4 for at least six months.
Goldman Sachs: 300 Million Jobs Exposed to AI
Goldman Sachs economists estimated generative AI could expose 300 million full-time jobs to automation and raise global GDP by 7 percent.
AutoGPT and the autonomous-agent craze
AutoGPT wrapped GPT-4 in a self-prompting loop in March 2023 and became one of GitHub's fastest-rising projects ever.
Italy's Garante temporarily bans ChatGPT over data protection
On March 31, 2023 Italy's data protection authority ordered a temporary block on ChatGPT, the first such action by a Western regulator.
gptme launches as early open-source terminal coding agent
gptme emerged as one of the earliest open-source terminal AI coding agents, combining file editing, shell execution, and web browsing in a persistent conversation interface.
Sal Khan's TED talk argues AI could save education
At TED2023 Sal Khan argued AI tutors could finally solve Bloom's two-sigma problem and demoed Khan Academy's Khanmigo tutor.
BabyAGI, the task-driven autonomous agent
Yohei Nakajima's BabyAGI showed in April 2023 that a task-planning AI agent could fit in about a hundred lines of code.