AI Writes Over 25% of Code at Google

                                                                              Photo by Google DeepMind on Unsplash

AI has been cranking out over a quarter of Google’s code and have developed their own internal tool to help developers deliver. That is according to Google CEO Sundar Pichai during Alphabets earning call last Tuesday.

Wait. It’s not yet time to retire your coding career, but this does represent a major change in our industry and does require attention from software developers the world over.

Welcome To The Future

Google’s internal AI (Goose) can do more than Google’s Gemini.

When put to task in Google it has the ability to understand Google’s internal proprietary systems. Not only does it produce new code, but it also takes commands in natural language to modify or update existing code.

With Google using AI to generate over 25% of production code we are seeing a move from simply assisting developers to a point where it is actually producing code.

To some extent this is a boon to productivity, which may be no accident since Google have recently lowered headcount and still need to deliver features. Yet more ominous questions lie beneath the surfaces, those about the future of junior roles, the stability of tech careers, and what it really means to be a “developer.”

The Danger of Outsourcing Basic Coding

Jack Dorsey (former Twitter boss) made a prediction in 2020. That prediction was that junior roles will become a thing of the past.

“A lot of the goals of machine learning and deep learning is to write the software itself over time so a lot of entry-level programming jobs will just not be as relevant anymore”

If AI tools write beginner-level code with accuracy, what room is there for new programmers when their traditional work — the slog through the routine — is now handled by the machines?

We are destroying the pipeline of coders. The industry risks a future where there is no talent to become the senior developers of the future, meaning there will be no-one to oversee the AI-written code of tomorrow.

Ironically, by making today’s programming faster, AI could slow down the future of development — there’s no training data for experience.

The Role of Developers from 2024

The role of software developer is changing fast, as software developers change into AI supervisors rather than coders. As we move into the future more of our time will be spent managing AI outputs and directing AI models.

It’s like moving from the job of a builder to an architect. The work is not only more strategic but needs different skills. Even for basic coding tasks a well-trained developer needs to check and polish code, and that grunt work is something Sundar missed in their announcement.

It’s Time Software Developers Learned AI

Companies are no longer just looking for developers who know Python or JavaScript. Instead, they’re after those who can guide, interpret, and correct AI-generated outputs. According to a 2024 Microsoft and LinkedIn report, 66% of employers would hire a less-experienced developer with AI skills over a seasoned programmer who lacks them. That means if you’re planning to hang onto your career in software, it’s time to get comfortable with the AI toolkit and fast.

So, don’t panic, get used to the technology and use it to help you deliver. Just don’t leave it too long.

Developers Still Rule

Despite this news there is still reason to be skeptical of the AI gold rush. Every developer will not become a machine learning expert. The current technology is far from foolproof, so while getting in on the basics is important, we still have time until AI even attempts to replace the deep technical expertise of experienced programmers.

While AI might be able to write code, there is still opportunity for that human touch. AI lacks situational awareness, strategic insight or ethical compass. In my experience as we give more complex problems to AI, the nuances of the challenge are lost and the solutions fail to deliver. 

AI can’t even self-test, self-debug or self-correct. It’s developers who need to check the output, and because of that we’ll stay employed, at least for now.

Conclusion

Playing with AI is still fun for the curious. Yet as the machines start to threaten job stability it should be met with cautious skepticism rather than outright fear.

Right now, junior developers are at risk, and more senior roles are likely to change. As ever with software development, if you wish to keep working, you’ll need to change with the times.