AI Begins to Take Over Coding, Programmer Roles Shift

The profession of programmer is synonymous with the activity of writing lines of code (coding). However, in the era of artificial intelligence (AI), that busyness is beginning to change.

Now, most coding work is starting to be taken over by AI. Perplexity AI CEO Aravind Srinivas highlighted this shift.

He expressed his view by reposting a social media upload discussing how large language models (LLMs) are beginning to automate coding work.

The post, which garnered more than 15,000 likes and nearly one million views, also stated that the impact of AI automation is bringing computer science back to its roots closer to mathematics and physics.

In response, Srinivas briefly commented “well said”. His response seems to indicate that the view aligns with what is now beginning to happen in the industry.

Meanwhile, Anthropic’s “AI Exposure” index places programmers at around 75 percent coverage of tasks by LLMs, the highest compared to other professions they studied.

This data shows that AI not only makes the coding process faster but can also handle most tasks in that work.

Nevertheless, the changes occurring do not stop at efficiency alone. Previously, programmers spent much of their time writing and tidying up code, but now those tasks are starting to be automated by AI.

From having to assemble basic code structures first, fixing syntax errors, to generating code snippets, all of that is now replaced and done with the help of AI models.

As a result, the focus of programmers’ or software engineers’ work is also shifting. They are no longer just focused on writing code, but more on thinking about more complex matters.

For example, how systems can fail, design compromises (trade-offs), and ensuring architecture remains stable even when used at large scales.

This situation gives rise to the view that computer science is slowly returning to its original roots, namely logic, mathematics, and systematic thinking.

A similar opinion was also expressed by Anthropic CEO Dario Amode, who said that AI has the potential to handle most software engineer work in the next 6 to 12 months.

In fact, Amode said, some engineers at Anthropic now no longer write code directly. That means their main tasks have been handed over to be done by AI models.