Software engineers working in an open office space
getty
AI has changed coding forever. Just last month, Sam Altman, cofounder and CEO of OpenAI, posted a thank-you note to developers on X marking a massive shift in the industry.
“I have so much gratitude to people who wrote extremely complex software character-by-character. It already feels difficult to remember how much effort it really took. Thank you for getting us to this point," the post said.
The implication of the post was that the days of manual coding were coming to an end. A new generation of AI-driven tools like Claude Code, Cursor and Codex are now capable of generating code from scratch. Now, many up and coming developers are seeing less demand for their skills.
According to a Federal Reserve Bank of New York study released in 2025, fresh computer science graduates are facing unemployment rates of 6.1% to 7.5%, more than double that of biology and art history majors. As AI begins to automate code creation, many are starting to ask, is software engineering “cooked”?
The future of software engineering has been a frequent topic of discussion among AI leaders. In February 2026, Boris Cherney, creator and head of Anthropic’s Claude Code, said that the software engineer title is going to start going away by the end of the year and “it’s going to be painful for a lot of people.”
His comments align with Dario Amodei, cofounder and CEO of Anthropic who predicted at the World Economic Forum that “we might be 6-12 months away from models doing all of what software engineers do end-to-end.”
While Cherney and Amodei’s comments could be dismissed as an attempt to hype code creation tools, there seems to be a growing number of software engineers struggling to find work in the market.
Garrett Rose, a full-stack engineer with three and a half years of experience in the industry, recently put out a YouTube video claiming that the stress of the job market had contributed to a subconjunctival hemorrhage. Rose was laid off from his previous job 8 to 9 months ago, a layoff which he attributes to AI. Rose is now one of many engineers struggling to find roles in the industry.
“I am indeed very concerned with the future of software engineering. I understand that technology shifts and becomes more abstract from its core implementations. However, what AI is doing is training off of every software engineer that ever wrote code pushed to the web,” Rose told me via email.
While Rose doesn’t believe software engineering has been made obsolete by AI yet, he believes it could be in the future. He also said that AI’s ability to complete simple work has led to companies only hiring “unicorns," experienced developers with the flashiest resumes.
“Right now, software engineers have transitioned from writing code manually on a keyboard a couple of years ago to now speaking to agentic AI, describing exactly what they want to build and how to build it. The engineering part has only become the planning phase, and more or less everyone is an architect,” Rose said.
As developers across the market feel the pressure of automation, many companies continue to go full steam ahead with AI. Nowhere is this more clearly demonstrated than with the tokenmaxxing trend, where engineering teams are burning through as many tokens as possible to streamline product innovation.
However, though many software developers are being displaced by AI, a number of industry leaders suggest that engineers still have a critical role to play. Peter Hill, CTO at Synthesia, an AI video platform valued at $4 billion, told me via email that he didn’t believe software engineers had become obsolete.
“No, and I think people saying that are confusing the job with one specific part of the job. For a long time, the value of a software engineer was heavily tied to their ability to translate ideas into working code. That part is getting dramatically cheaper. But the reason companies needed that code in the first place hasn’t changed. Someone still has to figure out the right problem, what a great solution looks like, and when something is ready to put in front of a user,” Hill said.
Hill noted that AI hadn’t just changed the speed of writing code but also inverted the relationship between planning and building. Synthesia can now prototype in hours what used to take weeks with tools like Claude Code and Cursor. For instance, at a recent company hackathon, one team of coders managed to rebuild the company’s video editor from scratch in just a couple of days.
He said that Synthesia has been using tools like Claude Code, Cursor and other AI coding tools with promising results, such as faster feature work and easier automation of routine tasks. Hill also told me that at a recent company hackathon, the winning team rebuilt the company’s video editor from scratch in a couple of days.
When asked if he would look at manual coding tests when hiring new developers, Hill said, “I think the honest answer is that pure coding assessments are becoming less useful as a signal.” Instead, he’d rather see how someone thinks through a complex product problem and uses the tools available to them to respond.
Synthesia isn’t alone in turning to AI to automate code. Nancy Wang, CTO of password management giant 1Password, told me the company polled some of its best engineers to ask how much code they were writing manually and found that the top 10% were writing zero percent of code manually.
“I actually made the decision to pull manual coding tests from our software engineering hiring loop,” Wang told me in a video interview. “I want to see how you’re thinking about things, less so can you just generate a ton of code? Because guess what? LLMs are going to be way more effective at just generating pure lines of code than a human engineer will.”
In a world where writing code manually is becoming less important, Wang says developers need a mix of the systems-level thinking that would come with a traditional computer science curriculum, plus AI proficiency. “We’re moving into fully agentic workflows, and in order to build the right workflows, you need to know where the systems are bottlenecked and where different workflows might fail,” Wang said.
“At 1Password, we are requiring, you know, AI proficiency and AI fluency for all incoming engineers, as in, we will not hire anyone who is not at least proficient with AI tooling. And if they’re only proficient, that means they need to be also proficient and have depth in another area, whether it be databases, whether it be cryptography and so on, so fourth,” Wang said.
Fluency with AI tools isn’t just about knowing how to use Claude Code, but being able to create multi-step workflows and capture the state of multiple systems. Wang would also look for engineers with nuanced approaches to automation, such as using Claude Code to generate code and designs and then Codex to spin up test harnesses.
As the product development lifecycle bleeds into the software development lifecycle, Wang points to the birth of a product engineer role, someone who builds prototypes, has the ability to talk to customers and can step into the shoes of a product manager.
The rising popularity of tools like Claude Code has changed coding workflows for good. As companies like Synthesia and 1Password look to automation to streamline product innovation, developers will be under greater pressure to demonstrate proficiency with AI tools.
In the short term, developers will play a critical role by working alongside AI agents, double-checking code, and developing complex workflows that connect to downstream systems. But the future of the software engineer’s role is uncertain. If these tools become advanced enough to write and deploy code independent of human supervision, there will also be less demand for developers.
Software engineering is “cooked” in the sense that developers are going to have to work much harder to demonstrate their value in the market. If engineering teams can do more work with AI agents and less overall headcount, there will be fewer roles available to human engineers.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。