This article is crossposted from IEEE Spectrum’s careers newsletter. Sign up now to get insider tips, expert advice, and practical strategies, written in partnership with tech career development company Taro and delivered to your inbox for free!
Cursor, the AI-native code editor, recently reported that it writes nearly a billion lines of code daily. That’s one billion lines of production-grade code accepted by users every single day. If we generously assume that a strong engineer writes a thousand lines of code in a day, Cursor is doing the equivalent work of a million developers. (For context, while working at Pinterest and Meta, I’d typically write less than 100 lines of code per day.)
There are only about 25 million software developers worldwide! Naively, it appears that Cursor is making a meaningful percentage of coders obsolete.
This begs the question: Is it even worth learning to code anymore?
The answer is a resounding “yes.” The above fear-based analysis of Cursor misses several important points.
First, the demand for software is insatiable. Getting a computer to follow our instructions properly (i.e., coding) is economically incredibly valuable. However, for decades now, the number of people with the specialized skill to communicate with computers has been limited. If AI tools let us write more functional code, many more people, companies, and industries will benefit. As Marc Andreesen famously alluded to in his 2011 essay “Why Software Is Eating the World,” the demand for software is elastic. As software becomes cheaper, we see a disproportionately large increase in demand.
There’s no such thing as finishing a software project. Humans are very good at demanding more products and services, and software will inevitably play a role in that future.
Another critical observation is that AI-assisted code is simply a continuation of the evolution of software development. In the 1970s, we used to program with punch cards. Then we started writing low-level code with a keyboard. Then we shifted to higher-level languages. Each successive era saw large increases in the number of software engineers while simultaneously improving the productivity of each person. We’ve seen this storyline before: AI is just another tool to give us superpowers.
Coding tools like Cursor or Devin are less about replacing software engineers than they are about increasing the number of software engineers. Going forward, writing code won’t be restricted to those with a particular title; everyone including lawyers, accountants, and doctors will develop software. Cursor won’t replace the 25 million software engineers we have today. Instead, it will create millions more.
—Rahul
IEEE Spectrum’s annual ranking of the top programming languages is here—and it might be the last. Coders are now turning away from many of the public expressions of interest in different languages (like Google searches and Stack Exchange queries), which have served as key metrics for the ranking since 2013. What does the move toward AI coding tools mean for the future of programming and how we track the coding landscape?
Also in the realm of how AI is affecting programming: The Stanford Digital Economy Lab has released a report finding early evidence that employment has taken a hit for young workers in the occupations that use generative AI the most, including software engineering. The report used data from the largest payroll provider in the United States, ADP, to gain up-to-date employment and earning data for millions of workers.
The Trump administration recently issued an executive order adding a $100,000 fee to new H-1B visa petitions. Major tech firms like Amazon are among the largest H-1B employers, but some higher education positions may also be affected. Inside Higher Ed reports on which institutions have the most of these visas and how they compare to other industries.
From Your Site Articles
Related Articles Around the Web
