Who owns OpenAI?
OpenAI is an artificial intelligence (AI) research laboratory consisting of the for-profit corporation OpenAI LP and its parent company, the non-profit OpenAI Inc. The company conducts research in the field of AI with the stated goal of promoting and developing friendly AI in a way that benefits humanity as a whole.
The organization was founded in San Francisco in late 2015 by Sam Altman, Elon Musk, and others, who collectively pledged US$1 billion. Musk resigned from the board in February 2018 but remained a donor. In 2019, OpenAI LP received a US$1 billion investment from Microsoft and Matthew Brown Companies. OpenAI is headquartered at the Pioneer Building in Mission District, San Francisco.
Is OpenAI a private company?
In 2019, OpenAI became a for-profit company called OpenAI LP to secure additional funding while staying controlled by a non-profit called OpenAI Inc in a structure that OpenAI calls “capped-profit”, having previously been a 501(c)(3) nonprofit organization.
Who has invested in OpenAI?
AP Ventures and Mitsubishi Corporation co-led the round and were joined by Finindus, Nippon Steel Trading, Hillhouse Investment, Trustbridge Partners, SINTEF Ventures, and Firda. – 40Seas, a Tel Aviv-based fintech platform for cross-border trade financing, raised $11 million in seed funding
Does Elon own OpenAI?
OpenAI is an independent organization, and while Elon Musk has been involved with the organization in the past, he is not directly involved in its day-to-day operations or decision-making.
What are the products released by OpenAI?
Gym
Gym aims to provide an easy to set up, general-intelligence benchmark with a wide variety of different environments—somewhat akin to, but broader than, the ImageNet Large Scale Visual Recognition Challenge used in supervised learning research—and that hopes to standardize the way in which environments are defined in AI research publications, so that published research becomes more easily reproducible. The project claims to provide the user with a simple interface. As of June 2017, Gym can only be used with Python. As of September 2017, the Gym documentation site was not maintained, and active work focused instead on its GitHub page.
RoboSumo[edit]
In “RoboSumo”, virtual humanoid “metalearning” robots initially lack knowledge of how to even walk, and are given the goals of learning to move around, and pushing the opposing agent out of the ring.
Through this adversarial learning process, the agents learn how to adapt to changing conditions; when an agent is then removed from this virtual environment and placed in a new virtual environment with high winds, the agent braces to remain upright, suggesting it had learned how to balance in a generalized way.
OpenAI’s Igor Mordatch argues that competition between agents can create an intelligence “arms race” that can increase an agent’s ability to function, even outside the context of the competition.
Debate Game[edit]
In 2018, OpenAI launched the Debate Game, which teaches machines to debate toy problems in front of a human judge. The purpose is to research whether such an approach may assist in auditing AI decisions and in developing explainable AI.
Dactyl[edit]
Dactyl uses machine learning to train a Shadow Hand, a human-like robot hand, to manipulate physical objects. It learns entirely in simulation using the same RL algorithms and training code as OpenAI Five. OpenAI tackled the object orientation problem by using domain randomization, a simulation approach which exposes the learner to a variety of experiences rather than trying to fit to reality.
The set-up for Dactyl, aside from having motion tracking cameras, also has RGB cameras to allow the robot to manipulate an arbitrary object by seeing it. In 2018, OpenAI showed that the system was able to manipulate a cube and an octagonal prism.
In 2019, OpenAI demonstrated that Dactyl could solve a Rubik’s Cube. The robot was able to solve the puzzle 60% of the time. Objects like the Rubik’s Cube introduce complex physics that is harder to model.
OpenAI solved this by improving the robustness of Dactyl to perturbations; they employed a technique called Automatic Domain Randomization (ADR), a simulation approach where progressively more difficult environments are endlessly generated.
ADR differs from manual domain randomization by not needing there to be a human to specify randomization ranges
GPT-3
Main article: GPT-3
Generative Pre-trained Transformer 3, commonly known by its abbreviated form GPT-3, is an unsupervised transformer language model and the successor to GPT-2. It was first described in May 2020.
OpenAI stated that full version of GPT-3 contains 175 billion parameters, two orders of magnitude larger than the 1.5 billion parameters in the full version of GPT-2 (although GPT-3 models with as few as 125 million parameters were also trained).
OpenAI stated that GPT-3 succeeds at certain “meta-learning” tasks. It can generalize the purpose of a single input-output pair. The paper gives an example of translation and cross-linguistic transfer learning between English and Romanian, and between English and German.
GPT-3 dramatically improved benchmark results over GPT-2. OpenAI cautioned that such scaling up of language models could be approaching or encountering the fundamental capability limitations of predictive language models.
Pre-training GPT-3 required several thousand petaflop/s-days of compute, compared to tens of petaflop/s-days for the full GPT-2 model.
Like that of its predecessor, GPT-3’s fully trained model was not immediately released to the public on the grounds of possible abuse, though OpenAI planned to allow access through a paid cloud API after a two-month free private beta that began in June 2020.
On September 23, 2020, GPT-3 was licensed exclusively to Microsoft.
ChatGPT
Main article: ChatGPT
ChatGPT is an artificial intelligence tool that provides a conversational interface that allows you to ask questions in natural language. The system then responds with an answer within seconds. ChatGPT was launched in November 2022 and reached 1 million users only 5 days after its initial launch.
Music
OpenAI’s MuseNet (2019) is a deep neural net trained to predict subsequent musical notes in MIDI music files. It can generate songs with ten different instruments in fifteen different styles. According to The Verge, a song generated by MuseNet tends to start reasonably but then fall into chaos the longer it plays.
OpenAI’s Jukebox (2020) is an open-sourced algorithm to generate music with vocals. After training on 1.2 million samples, the system accepts a genre, artist, and a snippet of lyrics and outputs song samples. OpenAI stated the songs “show local musical coherence, follow traditional chord patterns” but acknowledged that the songs lack “familiar larger musical structures such as choruses that repeat” and that “there is a significant gap” between Jukebox and human-generated music. The Verge stated “It’s technologically impressive, even if the results sound like mushy versions of songs that might feel familiar”, while Business Insider stated “surprisingly, some of the resulting songs are catchy and sound legitimate”.
Whisper
Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification.
API
In June 2020, OpenAI announced a multi-purpose API which it said was “for accessing new AI models developed by OpenAI” to let developers call on it for “any English language AI task.”