Artificial Intelligence(A.I)
Areas of application of artificial intelligence
Last updated
Areas of application of artificial intelligence
Last updated
The main developers of artificial intelligence systems
Business leaders like Google and Amazon, as well as smaller market players, are also exploring the possibilities of artificial intelligence. Even companies unknown to most, such as BotsCrew, which develops chatbots for Telegram and other services, are making great progress in terms of AI development. And companies like InData Labs and Nexocode create programs for analytical tasks, data processing, and statistical reports on requests from customer companies. We present just a few companies that are major players in this field, and whose developments will soon change our lives.
Microsoft started building a new supercomputer in its Azure cloud storage in 2020, working together with OpenAI. The aim is to train algorithms and enhance the functionality of AI by connecting relevant infrastructure components for use by other organizations and individual specialists. Additionally, the company has released Microsoft Designer, a program designed for designers who create graphic images for social networks and other resources.
Alphabet. The operations of Google and YouTube, the parent company, heavily rely on the use of AI. Automated algorithms are utilized to compute advertising expenses, promote content, and set up spam filters in Gmail. Alphabet's subsidiaries, DeepMind and Waymo, are also keeping pace. DeepMind develops its own AI software, while Waymo produces self-driving vehicles. Waymo is recognized for launching a completely driverless taxi service that is adapted to public roads in 2020.
Amazon. has fully integrated AI into its structure and uses it to create targeted advertising, search engines, and Amazon Web Services. Many American families use Amazon Alexa today. For AWS cloud clients, the company offers various AI-related services, such as advanced text analytics, automated code checking, and chatbots.
Nvidia. produces powerful and high-speed chips that can be used to run the most complex AI applications. The Leonardo supercomputer, which is among the fastest, is powered by Nvidia GPUs. Many large companies that don't have their own supercomputer use equipment with built-in Nvidia chips and Nvidia Quantum InfiniBand software.
IntuitiveSurgicall's da Vinci Surgical System uses advanced surgical automation and imaging technology to enable minimally invasive operations. Today, the company's experts are working on programs that can process even more specialized information to become full-fledged assistants during surgical procedures.
IBM. Despite its extensive experience in building computer systems, IBM continues to conduct research in the field of AI development. IBM is a leader among companies offering artificial intelligence technology, and customers use its AutoML and AutoAI products to create and train AI models.
OpenAI is one of the leading companies in the field of AI, and its technology is no longer seen as something seen only in Sci-Fi. People are now using AI to write their dissertations and even books. In this article, we will tell the story of OpenAI's formation and highlight some of its most successful projects.
Getting Started
The nonprofit OpenAI Lab was founded in San Francisco, California, on December 11, 2015. It was led by a number of investors - Elon Musk, Sam Altman (former president of the IT business incubator Y Combinator) and other businessmen from Silicon Valley. Investors have pledged $1 billion to develop programmer-friendly artificial intelligence systems that would "benefit all of humanity.
Earlier Elon Musk stated that AI could be more dangerous than nuclear bombs. The new company planned to release open-source software so that researchers could develop artificial intelligence жsystems in the future. The company also released a language modeling toolkit called OpenAI Codex and an AI safety research agenda.
The first was the Gym platform. It allowed training AI with games and algorithmic tasks. The machines made the decisions for which they received the highest rewards. This process can be compared to training a dog: the bot was rewarded every time it did a task correctly. The second product is Universe. This is a set of tools for training programs - intelligent agents. Universe allows artificial intelligence to use a computer like a human: a neural network analyzes information on the screen and works with a virtual mouse and keyboard. The product was created to train an AI to perform the same tasks as a human. Care Mask and Neural Networks for Text Generation The most famous name in the history of OpenAI is Elon Musk. The entrepreneur left the board of directors three years after the company was founded, in 2018, but continued to invest in the project and advise the team. The company explained this decision as follows: the team felt that in the future there could be a conflict of interest with Tesla, a development of Elon Musk. The company produces electric cars and is also engaged in the development and implementation of AI.
Musk himself wrote on Twitter that he left because he "disagreed with something about what the OpenAI team wanted to do. Perhaps it was related to the neural network for text generation presented in the same year. In 2018, OpenAI introduces the GPT algorithm. The task was to create a model of an NLP algorithm, that is, natural language processing. GPT used 5 GB of book texts to generate its own. In 2019 they refined it, added parameters to build clearer texts, and introduced the GPT-2 product. A neural network can translate publications, understand the logic of the text, finish sentences, and summarize. The database already contains 40 Gb of text, that is 7 thousand books and 8 million sites.
The algorithm became famous because it could make up fake news based on headlines . While this was one of the concerns raised by the OpenAI team, the primary reason for not releasing the model was to prevent it from being used maliciously to generate spam, phishing emails, and other forms of online abuse.. OpenAI decided not to release a trained model, fearing that the Internet could be flooded with fakes. The third version (GPT-3) was introduced in 2020. The algorithm has even more features - for example, it can write a real article for the media. The database contains 570 GB of text. The third generation model can solve almost any text tasks in English, the only question is the quality of the content.
In 2019, IT giant Microsoft invested $1 billion in the company, and OpenAI became a commercial organization. In 2021, the corporation also financially supported the project. And in January 2023, OpenAI and Microsoft signed a multi-year deal worth about $10 billion. The exact amount was not disclosed. Under the agreement, OpenAI can use Microsoft's infrastructure to develop and use its products, and the corporation receives the right to prioritize commercialization of new services.
Microsoft has explained what the cooperation will consist of:
companies are working in tandem to develop new supercomputing technologies;
Microsoft programmers have access to OpenAI developments and can use them to work on new AI technologies;
Microsoft Azure is the exclusive cloud provider for OpenAI.
The company unveiled its most discussed development to date - ChatGPT chatbot - in November 2022. In just the first five days the service was used by one million users. Chat-bot is based on an improved version of the GPT-3 algorithm - GPT-3.5, which was released in the same year 2022.
Artificial intelligence on demand can write poems, songs, scientific papers, movie scripts, content plans for social media accounts, and find errors in code. In addition, he can give advice and support dialogue. At the same time he is very correct and avoids controversial topics. In English ChatGPT handles tasks better than in Russian. The neural network is free for users, you only need to register on the company's website. There is also a paid version of the product.
What about the future?
The OpenAI team at the beginning of 2023 employed 375 people, which is how the startup has grown since its founding in 2015, when the young company had only nine employees. Today OpenAI is one of the most expensive startups in the United States, valued at $29 billion.
Experts believe that ChatGPT can change many areas. We already know that a student was able to defend his diploma written by a neural network. Now some authors are using the bot to gather facts for writing materials; programmers are trying to generate code. For bartenders, a neural network can offer recipes for new cocktails, for gym-goers it can offer programs for classes, and for teachers it can offer lesson plans.
You can discuss everything with the bot, except news. The algorithm was trained on the basis of data current only until October 2021. Now available version GPT-4 , and in winter 2023/24 promise to launch GPT-5
Neural networks still have many unexplored possibilities. For example, the generation of video content - this is exactly the kind of project that OpenAI is currently working on. It is planned to be presented in the near future.
Areas of application
The applications of artificial intelligence in society are expanding and becoming more diverse, but this does not mean that machines will completely replace humans. In fact, it is not necessary to strive for such a goal. The operation of any system, even a highly intelligent one, must be controlled and directed by a human. AI should be used to simplify routine tasks and replace humans in workplaces with difficult conditions. This shift is expected to occur in the coming decades.
Furthermore, it is impossible to hold a machine accountable for a mistake, even if it results in tragic consequences. For instance, if an unmanned car hits a person or a patient dies on the operating table during a procedure performed by a robotic surgeon, who should be held responsible?.