Microsoft's $10 billion bet on ChatGPT's developer marks a new era in artificial intelligence (AI)

Tech companies are trying to get into the new field of "generative artificial intelligence" as quickly as possible.

Microsoft is thinking about giving $10 billion to a San Francisco company called OpenAI that does research. This deal could be the most important in the history of AI.

If the US software giant is right about the technology's far-reaching effects, it could also cause a reorganisation in the AI world as other tech companies rush to find their place in the new field of generative AI.

Last month, when OpenAI released ChatGPT, it made headlines all over the world. ChatGPT is an artificial intelligence system that can answer questions and write natural-sounding text.

But Microsoft executives think that the service's technology will soon have a bigger impact on the tech world as a whole.

Eric Boyd, who is in charge of AI platforms at Microsoft, said that the way people use computers will change because of these AI models. He said that the way people use technology every day will change when they can talk to a computer like they would talk to a person.

Microsoft's possible investment, which was first reported by the newsletter Semafor last week and confirmed by two people who know about the situation, would give it a large minority stake in OpenAI. OpenAI would be worth $29 billion after the investment.

Venture capitalists are rushing to back the latest trend in AI at a time when other investment trends, like blockchain and cryptocurrencies, are losing popularity.

In 2019, Microsoft put $1 billion into OpenAI for the first time. This made OpenAI the tech platform for Microsoft's complicated AI models and gave Microsoft the first chance to sell its technology.

The software giant has already used OpenAI's technology in a number of its own products, but its executives say this is just the start of what's to come.

Since 2021, its cloud customers have been able to pay to use GPT-3, an AI model that creates text. Last year, image-making systems like Dall-E 2 caused a big stir in the AI world. It is the basis for a new Microsoft product called Designer and is also available through the Bing search engine.

In the meantime, GitHub, a Microsoft service for developers, has made a product out of Codex, a system that tells programmers what lines of code to write next.

Thomas Dohmke, the CEO of GitHub, says that the AI system writes 40% of the code that developers who use the Copilot service write. This cuts in half the time it takes to write new code, which is a huge improvement after a decade of mostly ineffective efforts to make developers more productive.

"It's a number about productivity that blows my mind," said Dohmke.

A lot of OpenAI's technology comes from what they call "large language models," which are trained on a lot of text. Unlike earlier forms of machine learning, which have been the most important in AI for the last ten years, this method has led to systems that can be used in more situations, which increases their commercial value.

"The real power of these models is that they can do so many different things at the same time," said Boyd at Microsoft. He also said that this makes "zero-shot" learning possible, which means that the AI can be used for new tasks without being trained.

Google and other tech giants, as well as a few start-ups, have also spent a lot of money and time making big AI models like this one. But since GPT3's ability to make large blocks of text on demand shocked the AI world in 2020, OpenAI has set the pace with a series of interesting public demonstrations.

Microsoft executives want to use this technology in a wide range of products. Satya Nadella, the CEO of Microsoft, said at a company event late last year that generative AI would lead to "a world where everyone, no matter what they do, can get help from technology for everything they do."

Oren Etzioni, a board member and advisor at A12, an AI research institute co-founded by Paul Allen, says that generative AI will play a big role in "productivity" apps like Microsoft Office.

He said that in the future, all workers will use productivity software that shows them relevant information, checks their work, and gives them the option to make their own content.

Microsoft's rivals have noticed that this could change the software industry. They see this technology as a rare chance to get into markets where big tech companies are already strong.

Emad Mostaque, the head of London-based Stability AI, which made a big splash last year when it released its open-source image-generation system, said that his company was building a "PowerPoint killer"—an AI tool that will make it much easier to make presentations than the widely used Microsoft application.

This is both a defensive and an offensive move for Microsoft, which wants to protect its well-known products like Office and fight harder in markets like internet search.

With a possible investment in OpenAI, Microsoft is also trying to position itself as the main platform on which the next generation of artificial intelligence will be built.

"The amount of cloud computing power [OpenAI] needs is too much for a start-up or venture capital investor to handle," said one of the company's investors. This person also said that OpenAI had no choice but to try to get money from one of a small number of tech giants.

With its first investment in OpenAI, Microsoft tried to get ahead of the competition by building what it calls a "supercomputer" to train the research company's large AI models. For its AI work, Meta, a Facebook-owned company, now uses the same technology platform.

Nadella said that working with OpenAI gave the company a head start, which meant that the company's AI supercomputer could do calculations for about half the price of its main competitors.

It could be important to find ways to save money. Analysts at Morgan Stanley say that using ChatGPT to answer a question costs about seven times as much as a normal internet search. This is because natural language processing is more expensive. The companies that compete with Microsoft the most in cloud computing have also been trying to work with some of the most promising companies in generative AI. But none of the other new companies in the field have made AI models that are as big or as wide as OpenAI.

Stability AI has a three-year contract with Amazon's cloud division. In 2021, three Google researchers started a company called CohereAI, which agreed to train its own AI using the search engine's computing platform.

Regulators might be worried if a few tech giants become the main platforms for and investors in the start-ups that are making the next generation of AI technology.

One person who knew about Microsoft's investment plans said that the company's partnership with OpenAI would likely be closely looked at, but that the minority investment shouldn't cause the government to step in.

But since Big Tech owns the cloud computing platforms that will be needed for the age of generative AI, it seems likely that they will have a big say in what happens next.