With a view to developing an Artificial Intelligence that can greatly and surely benefit humanity, OpenAI was created in 2015. This company is located in San Francisco and is owned by Elon Musk and Sam Altman. Contributions to creation would amount to one billion US dollars.
OpenAI and its human resources
Elon Musk and Sam Altman embarked on the creation of OpenAI along with other investors. The aim being of course to form an endowment ofa billion dollars. The company has also declared that it is open to collaborations with other institutions. With this in mind, patents and research have been opened to the public within the limits of respect for security.
Nine researchers participated in the start-up of the company in 2016. Two years later, one of the two main founders, Elon Musk, decides to resign due to conflict of interest vis-à-vis Tesla which is none other than his workplace. That said, he still remains a donor. The key people at OpenAI are therefore currently Sam Altman the CEO, Ilya Sutskever the research director, and Greg Brockman the CTO. The strength of the OpenAI group is based on its human capital. Indeed, the group has excellent researchers in its ranks.
What are the company’s objectives?
OpenAI’s main vocation is to work in the field of artificial intelligence. Moreover, the suffix ” HAVE “ indicates Artificial Intelligence. The founders as well as the collaborators considered that artificial intelligences involved risks. Indeed, too much use of AI has caused anomalies to arise. The main purpose of OpenAI is therefore to solve these problems. The ideal for the founders is the creation of an artificial intelligence endowed with greater safety and which can be more beneficial to the human being. For this to happen, the company is open to any collaboration, from any organization or any individual.
Focus on the accomplishments of OpenAI
In the field of artificial intelligence, OpenAI very quickly set up programs. Indeed, as of 2016, a still beta version of OpenAI Gym is available to the general public. The latter is a manifestation of a research base on learning. In 2020, OpenAI makes an announcement. It is a question of GTP-3 which consists of answering questions in a more natural human language. This one is formed with a lot of words taken from the internet. Moreover, it is also able to translate languages since an API is associated with it. Remember that the strength of GTP-3 lies in the facility to produce coherent texts, and this, in more than one language.
It should be noted that while OpenAI was originally a not-for-profit entity, it is now capped-for-profit. We will even notice a Microsoft – OpenAI association.
OpenAI has never stopped innovating for optimal development. Its products are diverse, and mostly focus on reinforcement learning.
Gym is a documentation site in order to provide the user with a simple interface. Indeed, it offers a general intelligence reference that is easy to configure with a wide variety of environments. In fact, its purpose is to standardize the definition of environments in AI research publications. This research will be easier to replicate. Gym is like a larger version ofImageNetthe large-scale visual recognition challenge.
As of June 2017, it can only be used with Python. Then in September 2017, the site was not maintained and active work instead focused on its GitHub page.
“RoboSumo” is to teach virtual humanoid robots ” metalearning” see you 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 considered that he had learned to generalized balance.
According to OpenAI’s Igor Mordatchcompetition between agents can create a “arms race” intelligence. So even out of competition, the agent’s ability to function could increase.
Debate Game was launched in 2018. Its purpose is actually to research an approach that can help audit AI decisions and develop explainable AI. In this debating game, the machines are introduced to debating toy issues in front of a human judge.
Using machine learning, dactyl form a robot Shadow Hand to start from nothing. However, he uses the same algorithm code thatOpenAI Five. The robot hand is fully trained in a physically inaccurate simulation.
Generative pre-training or Generative Pre-trained Transformer (GPT) demonstrates his abilities to gain knowledge of the world and deal with long-term addictions. Then come his successors GPT-2 and GPT-3from language models of transformer unsupervised. GPTs are like general purpose learners.
MuseNet is a deep neural network. It can generate songs with ten different instruments in fifteen different styles.
APIs consists in particular of « access new AI models. It can be called for “any AI task in English”
DALL-E and CLIP
SLAB is a Transformer model that creates images from textual descriptions. On the contrary, CLIP creates a description for a given image.
OpenAI Microscope was created to easily analyze features that form inside neural networks.
OpenAI Codex is a descendant of GPT-3. It is considered the AI powering the code autocompletion tool GitHub Copilot.
Video game robots and other benchmarks
OpenAI Five or five robots able to play against human beings in video games. This particular team of players is basically guided by trial and error algorithms. She plays on Dota 2 : the famous five-on-five fighting video game.
The first steps of OpenAI Five
It’s during The international 2017 – the first tournament of the game’s annual championship – that a professional player suffered his first defeat against a robot. Dendi, the player in question had fought live against a bot and had lost. Greg Brockman, the CTO, explained the situation: this victorious bot had been training against itself for two weeks. He was learning to play through a reinforcement learning mechanism. When he managed to achieve one or more objectives in the game, this robot was rewarded. Intensive training had taken place for several months for these particular players. Learning software aimed for advancement in programming proficient software to solve 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 was able to participate in The international 2018 : two matches against professional players however they came out losers.
But in April 2019 in San Francisco, they beat OG, the then-world champions 2-0. Towards the end of the month, the bots continued to play in a four-day online gaming competition. Their victory amounted to 99.4% out of a total of 42,179 games played.
Reinforcement learning on platforms is today a very popular field of research. Gym Retro is one of them. Generally, it is used to conduct research on RL algorithms and to study their generalization. Long before, research in RL was based on the optimization of agents in solving unique tasks. Gym Retro allows in the same way to make the generalization between games but by differentiating on appearances.
We wish to give thanks to the author of this short article for this outstanding material
OpenAI: all you need to know about Elon Musk’s AI business – LeBigData.fr
Visit our social media accounts along with other pages related to themhttps://www.ai-magazine.com/related-pages/