According to media reports on August 23, Replits, Anysphere, Magic and other companies have raised $433 million this year, bringing the total funding since the beginning of 2023 to $906 million, and a large amount of money has poured into AI programming assistants.

The Rise of AI Programming Assistants in Software Engineering

With the rapid development of artificial intelligence technology, AI programming assistants are gradually becoming the new favorite in the field of software engineering. This smart tool is designed to help programmers write code and be more productive. Since the beginning of last year, many companies have invested in "AI programming assistants", and the cumulative financing amount has approached $1 billion, indicating that computer programming has become the forefront of AI technology innovation.

Today, Silicon Valley is increasingly convinced of the huge potential of AI applications in the field of programming. Code.org CEO Hadi Partovi once quipped that software development without AI is now as obsolete as writing without word processing software. This view is widely circulated in Silicon Valley, although some investors are still skeptical about whether AI can bring real economic benefits.

An executive on the Code.org board even said that the company plans to stop hiring people who don't use AI programming tools by the end of the year. This move further highlights the importance of AI programming assistants in modern software development.

Investment firm Index Hannah, Partner at Ventures Seal notes that AI tools will be more likely to be commercially successful if they can be seamlessly integrated into existing workflows and show their value immediately. In the field of programming, the value of AI tools is obvious, so the future is very promising.

AI programming assistants are favored by Silicon Valley

Not only startups, but also tech giants like Microsoft, Amazon, Meta, and Google are racing to develop AI assistants that can help write and modify code. Code.org CEO Hadi Partovi points out that the easier programming becomes, the greater the need for technology, as it means that more technology can be developed.

As one of the world's largest software development platforms, GitHub, owned by Microsoft, was one of the first companies to turn large language models into programming assistants. One of the prototypes that GitHub initially developed eventually evolved into the name GitHub Copilot's AI programming assistant and was officially launched in 2022. Currently, GitHub Copilot has nearly 2 million paying subscribers.

GitHub CEO Thomas When they started using GPT-3, OpenAI's first major model, Dohmke said, they quickly discovered that their ability to write code was so good that they decided to build a product around it. Nowadays, GitHub The quality of code generated by Copilot has surpassed that of many average developers. As of April, GitHub's revenue grew 45% year-over-year, reaching $2 billion in annual revenue by the beginning of this month. Among them, Copilot contributed more than 40% to GitHub's revenue growth, and its business scale has exceeded the entire business volume when Microsoft acquired GitHub in 2018.

CEO of Microsoft Satya Nadella revealed that more than 77,000 organizations (including BBVA, FedEx, H&M, Infosys, and Paytm, among others) have adopted GitHub Copilot, a two-year-old tool, has seen a 180% year-on-year increase. However, despite this, IT departments at large companies still have concerns about the security of writing code in production using automated programming tools.

GitHub CEO Dohmke explains that AI-generated code is often manually reviewed before deployment. Some companies have reportedly seen productivity increases of 20 to 35 percent after using these tools, including Latin America e-commerce giant Mercado Libre and professional services firm Accenture. This shows that even with certain security concerns, AI programming assistants still show great potential to improve development efficiency.

AI tools are assistants, not substitutes for programmers

According to a McKinsey analysis last year, AI's direct impact on software engineering productivity could account for 20 to 45 percent of current annual spending, largely due to features such as generating initial code drafts, code corrections, and refactoring. McKinsey notes that generative AI is expected to enable software engineers to focus more on code and architecture design by speeding up the coding process.

Some software engineers have integrated AI assistants into their daily routines, not only increasing productivity but also stimulating creativity. For example, the CTO of Sereact, a robotics startup in Germany Marc Tuscher uses GitHub on a daily basis Copilot, and often in conjunction with ChatGPT to write code. GitHub Copilot is particularly useful for handling repetitive tasks, such as building user interfaces and back-end parts of a product.

Tuscher also uses ChatGPT to solve more abstract problems, such as proposing classic ideas or new research directions, and then asks how to implement those ideas in Python, and ChatGPT generates the corresponding code. He found both tools extremely useful.

Tuscher mentions that all the programmers he knows use these tools, and they really change the way they work. But he also stressed that AI tools are powerful assistants, not substitutes for programmers. Currently, generative AI does not yet have the skills to design a good software architecture or integrate systems, and these tasks still need to be done by programmers themselves.