Chip maker Nvidia to invest $50 million in drugmaker Recursion to accelerate the development of the biotech company's artificial intelligence models for the discovery of 
drugs. In the wake of the announcement, shares of Recursion rose by 80%. Shares of Nvidia also gained, posting more than 2% more.
AI: Nvidia invests $50 million in pharmaceutical company
Recursion (headquartered in Salt Lake City, Utah) uses AI-powered models to identify and design new therapies and offers those models to other drugmakers, including Roche and Bayer.
The company will use its biological and chemical datasets larger than 23,000 terabytes [editor's note: 1 terabyte (TB) equals 1,000 gigabytes (GB) or 1,000,000 megabytes (MB). 1 megabit, is 1 million bthet] to train his intelligence models 
artificial on Nvidia's cloud platform. These datasets are growing by “hundreds of terabytes every week,” Recursion CEO Chris Gibson said in a broadcast. AI models usually require large amounts of data, typically measured in terabytes, to train them.
Nvidia can then potentially license those AI models to BioNeMo, the company's cloud service for generative AI in drug discovery that launched earlier this year.
Nvidia's investment is the latest example of the focus on artificial intelligence making its way into the pharmaceutical industry. Drug makers are increasingly recognizing the potential of artificial intelligence to deliver life-saving treatments to people faster.
Moderna said in April it will harness the power of artificial intelligence to advance RNA technology, which is used in Covid vaccines. A month later, Google Cloud launched two new AI-powered tools that aim to help biotech and pharmaceutical companies accelerate drug discovery.
Nvidia, which makes chips used to power AI, hit a $1 trillion market value for the first time in June, part of the more than 200% increase in the stock price since the start of this year. year in the wake of strong demand for artificial intelligence.
Note:
In 2022 the Santa Clara company [Nvidia] launched the H100 processor. At first it seemed like a flop (it cost too much), then ChatGPT arrived. Currently most of the assembly activities for Nvidia are contracted out to Tsmc (Taiwan Semiconductor Manufacturing Co). According to CNBC, Nvidia accounts for 95% of the AI chip market share. “AI adoption is at an inflection point. ChatGpt by OpenAi has 
captured worldwide interest by enabling users to experience these applications firsthand and showcasing what is possible with generative AI,” he said. Nvidia CEO Jensen Huang. ChatGPT was released by OpenAI, a company funded by Microsoft and others since 2019. Who first extended the partnership (January 23), then announced the integration of the service into the Bing search engine and the Edge browser(February 7).
OpenAI is a non-profit artificial intelligence research organization with the aim of promoting and developing friendly artificial intelligence (AI) so that humanity can benefit from it. Some scientists, including Stephen Hawking and Stuart Russell , believed that if an advanced artificial intelligence gains the ability to redesign itself at an ever-faster rate, an unstoppable "intelligence explosion" would lead to the extinction of humanity. Elon Musk, one of the founders of OpenAI, defines AI as the "greatest threat to existence" of mankind. OpenAI has been structured as a non-profit so that the founders can focus on finding a long-term positive impact
The secret of Nvidia's success is the A100, a $10,000 chip that has become the true must-have of artificial intelligence professionals. The A100 is ideal for the type of learning models
automatic that feed tools such as ChatGp (specialized in man-machine dialogue). ChatGPT was launched on November 3, 2022 and has attracted attention for its detailed and articulate answers, although its accuracy has been criticized. Both supervised learning and reinforcement learning used human instructors to improve model performance.
For supervised learning it means that the machine is supplied with both input (i.e. of the data), both of the output, i.e. some examples that correspond to the expected result. In this way, through data processing, the machine must understand the connection between input and output and obtain a general rule to be applied thereafter.
TOreinforcement learning (or reinforced) is a machine learning method that is based on the definition of a "situation" and a "response" in a dynamic environment. In this case, the learning algorithm tries to learn to make decisions following a reward or punishment (i.e. the reinforcement) he receives as a result of the actions he performs. In practice, positive or negative feedback represented by a mathematical function of reinforcement. In other words, the learning system must learn to select the right action based on the situation it is in
ChatGPT is a chatbots based on artificial intelligence and machine learning developed by OpenAI specialized in conversation with a human user. The acronym GPT stands for Generative Pre-trained Transformer, a new technology applied to machine learning. Generative Artificial Intelligence, also called Generative AI or GenAI, uses Machine Learning and Deep Learning techniques to generate new data, including images, music and text, that did not exist previously.
chat bots, chatbots or chatterbot, is software designed to simulate a conversation with a human. The main purpose of these software is to simulate human behavior and are sometimes also referred to as intelligent agents.
GPT2 is an AI system that generates text via a tone and an argument. For example, when he was 
given the first sentence of George Orwell's novel "1984" as input, he produced a tale of a plausible future set in China. Unlike previous OpenAI products, GPT2 was not immediately released to the public to prevent illegal use, such as writing fake news
ChatGPT has been criticized because it often 'hallucinates', meaning that often invent his answers only to avoid admitting ignorance. Any user who has searched for an objective answer from ChatGPT knows this. A similar error caused a loss of 100 billion dollars to Google when they announced their embryonic AI software: Bard. Nvidia's solution to AI hallucinations is called «NeMo Guardrails». It is essentially another layer of artificial intelligence capable of «guide» the reference software. While ChatGPT uses its data for foresee which result will "spit out", NeMo would use them for check the correctness of the answer. In other words, ChatGPT will use the data to develop an answer, while NeMo will run the same data to check its plausibility. If NeMo does not find any matching data, the response will simply be «I don't know».
Related news: ChatGPT between benefits and potential risks, between reliability and unsolved problems
What is artificial intelligence
