Mark Zuckerberg Predicts AI Will Reshape Coding Forever in 2025

The end of software engineers?

Image made with DALL-E 3

Just a few days ago, Mark Zuckerberg said the following: "Probably in 2025, we at Meta are going to have an AI that can effectively function as a mid-level engineer that can write code. Over time a lot of the code in our apps will be built by AI engineers instead of people engineers."

The entire software industry is transforming before our eyes—not just the software being built, but also how it’s being built. A few weeks ago, Satya Nadella stated that applications will essentially evolve into agents, and now we’re hearing directly from Mark Zuckerberg that by 2025, they expect to have what is essentially a software engineer writing code for them. This claim is reasonable, considering that Meta has access to significant resources, both financial and in terms of GPU power, allowing them to effectively scale their team with AI-powered mid-level engineers.

All roads eventually lead to Rome

It’s worth noting that similar predictions have been made before. For example, NVIDIA’s CEO Jensen Huang made a comparable statement, and another example that comes to mind is Francois Chollet, the founder of the Ark AGI Prize and The Benchmark competition. When asked what might come after AGI, Chollet suggested that in the future, there will no longer be a need to write code manually; instead, you’ll simply specify what you want the program to do, and it will code itself.

This recurring idea is becoming more and more prominent in the industry, with an increasing number of voices suggesting that AI will have a profound impact on software engineering jobs.

Everyone seems to have a slightly different opinion about when we’ll reach that point, but the real question is: will AI replace people, or will it simply enhance their abilities to make them more effective? At the same time, we’re seeing growing signs that some companies may already be preparing for a future where they rely on fewer software engineers. For example, Salesforce has announced that it won’t be hiring additional engineers this year due to productivity improvements in their products driven by AI. Instead, the company plans to hire between 1,000 and 2,000 salespeople to help communicate the value of AI to their customers.

This doesn’t mean Salesforce is laying off its engineers—it just means that the employees they already have are now far more productive, and there’s no longer a need to fill those roles with new hires. In my view, this suggests that if you have a long career in software development and significant experience, you likely won’t be impacted by this shift. But for those just entering the industry or considering college to become software engineers, reading something like this could feel a bit intimidating or even discouraging.

As Francois Chollet pointed out, in the future, you might not need to be a technical user to create software. You could simply say, “Hey, I want an app that does this,” and if your request is vague, the AI would ask clarifying questions to better understand your needs and guide you through the process step by step.

In an interview, Zuckerberg also mentioned that an increasing amount of code will be written by AI, which will free us up to focus on more creative tasks. From my perspective, this suggests we’ll still need people to review the code. Initially, humans will handle this, but the speed at which projects move forward will likely depend on how quickly we can debug and refine what the AI produces. This means entire systems will need to be built to manage this new way of working.

In that sense, I agree that AI is more likely to complement humans than replace them. It will make us much more efficient. So rather than leading to mass job losses, a more realistic (and optimistic) scenario is that we’ll become significantly more productive.

The Promise of a New Future with AI

I have to admit that Zuckerberg’s vision for the future is incredibly optimistic (and I share that optimism). That said, I can’t help but ask: what happens when a software engineer is suddenly able to write 10, 100, or even 1,000 times more code than before? Does that mean they’ll have less work to do, or does it mean they’ll simply be far more productive? Personally, I believe it’s the latter.

We can see this reflected in broader technological trends. Whenever a technology becomes cheaper, faster, and more efficient, you might assume that companies adopting it would require less of it.

The reality is often the opposite.

As technology becomes more accessible, businesses and individuals tend to adopt it on a much larger scale. This happens because as the cost per unit of a technology decreases, the number of use cases in which it can deliver a positive return on investment increases exponentially.

That said, we need to be realistic—what we’re about to witness will be a complicated transition. There’s no denying that some people will lose their jobs, and there’s likely to be a period of friction as we figure out how to navigate this new reality. The software industry as a whole is changing at a breakneck pace, and we’re watching it unfold in real time, as reflected in the statements from tech leaders.

The idea that AI will simply eliminate coding jobs strikes me as an opportunity to create entirely new kinds of roles. Engineers will still be needed—people who can analyze problems and find creative ways to solve them. In the medium term, I think we’re heading toward a future where less hands-on coding is required, as more of that work gets delegated to AI.

Looking back at the past two decades, when we had to write every line of code manually, we’ve seen countless innovations that have made coding easier—like templates, autocorrect, and autocomplete. But even with these advancements, we’ve always needed the engineering mindset to understand problems, break them down into manageable steps, and figure out how to turn those steps into working code.

So, I find myself wondering: will these AI systems really be able to take on all—or at least most—of the work that engineers currently do?