With a view to developing Artificial Intelligence that can greatly and surely benefit humanity, OpenAI was founded in 2015. The company is based in San Francisco and has reportedly contributed US$1 billion to its creation.
OpenAI and its human resources
Elon Musk and Sam Altman set about creating OpenAI in the company of other investors. The aim, of course, is to to build a billion-dollar endowment. The company has also declared that it is open to collaborations with other institutions. With this in mind, patents and research have been opened up to the public, subject to security concerns.
Nine researchers were involved in the company’s start-up in 2016. Two years later, one of the two main founders, Elon Musk, decided to resign due to a conflict of interest with Tesla, his workplace. That said, he remains a donor.
In addition, then the evening of Friday November 17, 2023, the Board of Directors of OpenAI fired CEO and co-founder Sam Altman. The same board appointed Emmet Shear as CEOwho co-founded the AI star company Twitch. Several researchers have left the company since Sam Altman was dismissed in Greg Brockman in particular. The latter chaired the company and the board of directors
The key people at OpenAI are currently Sam Altman, CEO, Ilya Sutskever, Director of Research. The strength of the OpenAI group lies in its human capital. In fact, the group has some excellent researchers in its ranks.
A brief history
In 2019, Open AI becomes a for-profit company, having previously been a not-for-profit organization. The company allocates equity to its employees and enters into a partnership with Microsoft Corporation.
In June 2020, OpenAI presents GPT-3, a language model formed from trillions of words from the Internet. This model can be used to answer questions about natural language. In addition, it manages translation and systematically produces texts. called “improvised”.
Author of chatGPT
OpenAI a developed ChatGPT and launched it on November 30, 2022. ChatGPT is a natural language processing tool based on AI technology. It allows you to have human-like conversations and much more with the chatbot. The model linguistics can help you with a number of tasks such as writing e-mails, essays and codes.
OpenAI presented Dall.E in 2021, followed a year later by Dall.E-2, an algorithm generating realistic, accurate images at four times the resolution. The company has also created Whisper, an automatic speech recognition system.
What are the company’s objectives?
OpenAI’s main vocation is to work in the field of artificial intelligence. In fact, the suffix ” AI “indicates Artificial Intelligence. Both founders and collaborators considered artificial intelligence to be risky. Indeed, if AIs are used too extensively, anomalies can occur.
The main aim of AI is therefore toalleviate these problems. For the founders, their ideal is the creation of an artificial intelligence with greater security and which can be more beneficial to human beings. To achieve this, the company is open to collaboration from any organization or individual.
Focus on OpenAI’s achievements
OpenAI products
OpenAI has never ceased to innovate with a view to optimal development. Its products are diverse, focusing mainly on reinforcement learning.
Gym
Gym is a documentation site designed to provide the user with interface. Indeed, it offers a general intelligence reference that is easy to configure with a wide variety of environments. It also aims to standardize the definition of environments in AI research publications. This will make such research easier to reproduce. Gym looks like a larger version of ImageNetthe challenge of large-scale visual recognition.
Since June 2017, it can only be used with Python, but as of September 2017, the site has not been maintained and active work has instead focused on its GitHub page.
RoboSumo
“RoboSumo” consists of teaching virtual humanoid robots ” metalearning” to move by pushing the opposing agent out of the ring. With this learning process, agents learn to adapt to changing conditions. When moving to a more violent virtual environment, for example, the agent prepares to remain standing. It is assumed that it had learned to generalized balancing.
According to Igor Mordatch of OpenAIcompetition between agents can create a “arms race of intelligence. Thus, even without competition, the agent’s ability to function could increase.
Debate game
Debate Game was launched in 2018. Its purpose is to research an approach that can help audit AI decisions and develop explainable AI.. In this debate game, machines are trained to debate toy problems in front of a human judge.
Cocksfoot
Using machine learning, Dactyl forms a robot Shadow Hand at from scratch. However, it uses the same algorithm code asOpenAI Five. The robot hand is fully trained in a physically inaccurate simulation.
GPT
Generative pre-training or Generative Pre-trained Transformer (GPT) testifies to its ability to acquire knowledge of the world and deal with long-term dependencies. Next in line are its successors GPT-2 and GPT-3and language models of transformer unsupervised. GPTs are like general-purpose learners.
Music
MuseNet is a deep neural network. It can generate songs with ten different instruments in fifteen different styles.
It uses the same general-purpose unsupervised technology as GPT-2, a large-scale transformer model trained to predict the next token in a sequence, whether audio or text. The model is trained on data from MIDI files and can generate samples in a chosen style, starting with a prompt.
API
API includes “accessing new AI models. ” It can be called for “any AI task in English.”
DALL-E and CLIP
DALL-E is a Transformer model that creates images from text descriptions. On the contrary, CLIP creates a description for a given image.
Microscope
OpenAI Microscope was created to easily analyze features that form inside neural networks.
Codex
OpenAI Codex is a descendant of GPT-3. It is considered the AI powering the autocomplete code GitHub Copilot tool.
Video game robots and other benchmarks
OpenAI Five
OpenAI Five or five robots capable of playing against humans in video games. This particular team of players is essentially guided by trial-and-error algorithms. It plays on Dota 2 the famous five-on-five fighting video game.
The first steps of OpenAI Five
It was during The international 2017 – the first tournament of the game’s annual championship – that a professional player suffered his first defeat against a bot. Dendi, the player in question had fought live against a bot and lost. Greg BrockmanCTO, explained the situation: this victorious bot had been training against itself for two weeks. He was learning to play using a reinforcement learning mechanism. When it succeeded in achieving one or more objectives in the game, the robot was rewarded. Intensive training had been taking place for several months for these special players. The aim of the learning software was to advance the programming of software capable of solving complex tasks.
OpenAI Five between defeats and exploits
A year later, the bots could already unite in teams of five to play against amateurs and semi-professionals. OpenAI Five took part in The international 2018 two matches against professional players, although they came out on the losing side.
But in April 2019 in San Franciscothey had beaten OG, the world champions at the time, 2-0. Towards the end of the month, the bots continued to play during a four-day online gaming competition. Their victory amounted to 99,4 % out of a total of 42,179 games played.
Retro GYM
Platform-based reinforcement learning is a popular area of research today. Gym Retro is one of them. Generally, it is used to conduct research on RL algorithms and to study their generalization. In the past, RL research was based on optimizing agents to solve single tasks. In the same way, Gym Rétro enables generalization between games, but with a difference in appearance.
Projects developed by OpenAI?
One of OpenAI’s main projects is DALL-E, another AI model. DALL-E is a generative model capable of generating images from text descriptions. It ris based on the GPT-3 modelone of the most advanced language models developed by OpenAI.
DALL-E uses a transformer-based architecture, which enables it to process large amounts of data and generate high-quality images. It has been trained on a dataset of images and textual descriptions, enabling it to understand the relationships between the two.
OpenAI Five is also a project developed by OpenAI. This project has been designed to play the game in the same way as human players. It uses a combination of deep learning techniques and reinforcement learning.
This feature enables the model tolearn the strategic elements of the gameas well as the tactical skills needed to succeed. The model was able to defeat professional players in exhibition matches.
OpenAI unveils ChatGPT Enterprise
AI company OpenAI has launched ChatGPT Enterprisea generative AI chatbot service designed to meet the expectations of companies. The new version promises enterprise-level privacy and securityand unlimited access to the GPT-4 large language model (LLM). It also offers longer pop-up windows to handle longer entries, advanced data analysis capabilities and customization options.
ChatGPT Enterprise removes usage limits and is up to twice as fast. A 32k token pop-up window allows users to process entries or files four times longer. Unlimited access to advanced data analysis, formerly known as Code Interpreter, is also included. Advanced Data Analysis enables technical and non-technical teams toanalyze information in seconds. Scenarios range from the analysis of market data by financial researchers to the debugging of an ETL script by data scientists.