Anthropic, one of the world's leading artificial intelligence (AI) companies, is accelerating the process of replacing manual programming with automation based on AI.
According to Mike Krieger, Product Director of Anthropic, nearly 100% of the company's internal source code is currently created by Claude (artificial intelligence assistant), who is also the AI model developed by Anthropic.
Mr. Mike Krieger said that Anthropic's Claude AI system creates almost all of its own source code and products.
This assessment shows a major turning point in the software industry, when AI not only supports programming but also directly undertakes most of the development work.
Previously, Anthropic CEO - Dario Amodei predicted that 90% of programming code would be written by AI. However, according to Krieger, the actual rate has reached nearly 100%, far exceeding initial expectations.
This reflects the rapid development of AI programming tools, while raising a big question about the future role of human programmers.
According to the new model, humans no longer write each line of code, but switch to the role of monitoring, verifying and evaluating the quality of code created by machines.
Anthropic said they often process 2,000 - 3,000 code update requests simultaneously, especially in research groups with very rapid development speeds.
The company has built control frameworks and authentication procedures to ensure trust in the code generated by AI.
Not only writing code, Claude is also used to review and refute source code. Krieger said AI can detect security vulnerabilities, propose restructuring, and suggest optimal options.
“Claude was trained to become a good source code reviewer and has excellent critical thinking ability,” Mr. Krieger said.
These statements appeared just days after Anthropic's new workplace automation toolkit caused a strong sell-off of information technology stocks in India. Investors are concerned that AI is directly threatening traditional information technology service companies and SaaS software models, which are software provided in the form of online services.
Recently, Anthropic launched 11 new plugins for Claude Cowork, a virtual assistant designed to automate tasks in law, sales, marketing and data analysis.
The noteworthy point is that Claude assistants can now perform the work directly, instead of having to rely on platforms like Salesforce or ServiceNow.
This change makes many investors fear a "SaaS disaster", a term used by investment bank Jefferies to describe the risk of outdated traditional software companies.
The consequence quickly reflected in the market: shares of Infosys, TCS and HCLTech fell sharply, due to concerns that AI would reduce demand for outsourcing IT services.
Previously, at the World Economic Forum (WEF) in Davos, Anthropic CEO Dario Amodei continued to issue a noteworthy warning that the software engineer profession could become obsolete in the next 12 months. According to him, AI is shifting from a supporting role to automatically performing most of the work.
Emphasizing this view further, Sridhar Vembu, founder of Zoho (Indian technology group), recommends that the technology industry should seriously listen to Amodei, because Anthropic currently owns one of the most powerful AI programming tools in the world.
