What the latest Anthropic data says about software engineering jobs.
Finally a clearer picture about the role of AI in software engineering.
For years, the conversation around AI in software engineering has been dominated by one question: Will AI replace programmers? The business leaders who, still, seem to struggle to understand how engineers work, what makes them tick and how to measure their output ask questions like : Will it make development faster? More chaotic? Less creative? Now, thanks to new empirical research from Anthropic, we have hard data that shows exactly how AI is being used and perhaps a more relevant question is “is it making development easier for the engineers?”
Based on Anthropic’s data it’s clear that AI is not replacing engineers - it’s making them more efficient, and, in some ways, AI has already shifted the nature of the profession itself.
How will this play out? Will it lead to more engineering opportunities in the labor market, retaining higher wages, or will this lead to a commoditization of engineering, reducing salaries and opportunities? We debate both sides after a review of LLM usage.
AI is Fundamentally Changing Software Engineering
While there is very wide adoption of LLMs like ChatGPT and Claude, Anthropic’s study, 37% of Claude usage is related to software engineering tasks. Three times more than office and administrative use. It appears thus far to be the sector benefiting from AI adoption the most.

That’s not to say, however, that business leaders can take out 37% of headcount spend (though that may be the first thought). It turns out that AI augments an engineer far more than it automates one, making head-count-based ROI calculations a little harder.
Augmentation vs. Automation: AI’s Role in Engineering
The Anthropic report sets out a distinction between augmentation and automation in AI’s role within software engineering by splitting tasks into five categories: Directive, Feedback Loop, Task Iteration, Learning, and Validation.
When looked at through this lens, the report then assigned interactions into either bucket.

57% of AI interactions were augmentative - assisting developers to write, refine, and debug code (Centaur Model)
43% of AI interactions are automative - AI completed tasks with little human involvement or oversight. (Cyborg Model)
This shows that AI acts more as an enhancer than a replacer of engineers, being more effective at enhancing certain cognitive and analytical more than others.
So understanding how AI enhances rather than replaces engineering work requires us to look deeper at the specific skills that are amplified by AI.
What Skills Are Being Enhanced by AI?
The report validated some of the theory behind AI and found that it supports key cognitive tasks such as:
Critical thinking
Writing and documentation
Software programming
Complex problem-solving
Unsurprisingly, AI has minimal impact on manual and physical skills, such as hardware maintenance or network installation.

Who Is Using AI the Most?
The report goes deeper on the usage statistics provided earlier and shows that of the 37.2% of all recorded AI interactions, 16.8 percentage points was for developing and maintaining software.
Notably, users with roles like Computer Programmers and Software Developer accounted for less than a third of software queries. Does that mean we’re starting to see the rise of the Non-engineer who codes?
It’s now a reality that business analysts, product managers and even marketeers can become coders. I’ve used Replit’s AI agent to develop various custom tools (including our website, try the Tools section). I can safely say that the barrier to entry for engineering has almost fallen over. I feel moderately confident I could pull together a working MVP of most products I can imageine, without an engineer in the room. That’s helpful for me - faster time to market, lower costs etc - but what does this mean for engineers themselves?
This Anthropic study implies that a democratization of software development is underway. However, making code is one thing - maintaining and managing it is another, perhaps evolving role of professional software engineers in an AI-augmented landscape. We can now see how these changes are reshaping engineering roles.
Specializations Emerging in AI-Augmented Engineering
At Future Work AI, we conduct industry scenario planning exercises, including one that mapped out software engineer roles, ranging from the Traditional Code (no AI) through to the AI-First Software engineer. One of the striking things is how many coding roles are becoming hybrid positions - integrating AI-driven methodologies while taking on other responsibilities:
AI-Assisted Developer – Utilizing AI for code generation, debugging, and optimization.
Prompt Engineer – Specializing in refining AI-generated code outputs.
AI Ethics & Compliance Specialist – Ensuring responsible AI development practices.
AI-Driven DevOps Engineer – Automating infrastructure deployment and monitoring through AI.
At the same time, the Non-engineer who codes, like myself, lacking in a formal engineering education can now use low-code and no-code platforms to build applications without traditional programming expertise.
The big question: Will AI Reduce Software Engineering Jobs?
I’m not going to pretend to understand the future. AI is happening so fast, perhaps the acceleration of AI-generated custom apps will create a mountain of spaghetti code for traditional engineers to fix? Maybe instead it will automate enough that businesses will feel ready to choose between retaining developers and accelerating plans, or to maintain the same velocity for fewer heads.
Here’s both sides of the argument:
The Argument for Job Reduction
Entry-Level Jobs May Decline – AI can handle junior-level tasks such as writing boilerplate code, debugging, and generating documentation.
Companies May Shrink Engineering Teams – Increased efficiency could lead to fewer developers being needed for the same output.
Downward Pressure on Salaries – The democratization of coding through AI may increase competition, potentially lowering wages.
The Argument for Continued Job Growth
New AI-Related Roles Are Emerging – AI-assisted engineering positions, prompt engineers, and AI governance specialists are becoming more in demand.
AI Expands the Need for Skilled Oversight – While AI can generate code, it still requires human expertise to validate, optimize, and deploy at scale.
More Developers Means More High-Level Demand – As AI allows non-coders to create applications, experienced engineers will be needed to manage and refine complex AI-driven workflows.
Business leaders across the world have signalled that they intend to use AI to reduce their cost base. In fact the World Economic Forum reports that 41% of employers intend to downsize their workforce as AI automates certain tasks. Thus far most cuts seem to be in operations and customer service.
The picture for software engineers is less clear. Despite claims from Zuckerberg about mid-to-senior engineer cuts this year, these have yet to materialize in any meaningful way. The reality is probably in the middle - at least for now. AI will reshape engineering roles rather than replace them outright.
At the same time, lower barriers to entry and the democratization of coding through agentic workforce will make everyone a coder to some extent which is likely to put downward pressure on wages due to simple supply and demand considerations. Either way, there seems no future for the traditional coder, prioritizing full control over development speed by losing out on velocity.
The Future of Software Engineering is AI-Driven
The Anthropic report gives a clear message: AI is here to stay in software development, and its role will only expand. It presents opportunities and challenges for both engineering leaders and business leaders.
Priorities for organizations moving forward:
Balance AI-augmented development with robust quality standards
Invest in both traditional engineering and AI-specific capabilities
Develop frameworks for evaluating and integrating AI tools effectively
Prepare for a hybrid workforce where traditional engineers work alongside AI-empowered non-engineers
At Future Work AI, we research these evolving dynamics and their implications for technical organizations. Our ongoing analysis of AI's impact on software engineering continues to reveal new insights about this rapidly changing landscape.
Tell us below how your company is thinking about AI and what this means to you. If you have thoughts or perspectives on this topic, we’d love to hear it.
🙏🙏🙏