Will AI Replace Software Developers? 4 Senior Developers Weigh In

Artificial intelligence tools could be used to replace software engineers at major tech companies, according to some Silicon Valley CEOs. 

Developers have hit back at these claims, arguing that replacing lower-level positions and having more experienced devs oversee the work of AI could lead to detrimental results in the long term – even if there is a short-term boost. 

“AI Will Do the Work of a Mid-Level Engineer”

Speaking on the Joe Rogan Experience, Meta CEO Mark Zuckerberg said: “This year, we at Meta – as well as the other companies working on this [AI] – are going to have AI that can be a mid-level engineer and write code.

“In the beginning, it will be really expensive to run [to] get it to be more efficient. Then over time, a lot of the code in our apps including the AI that we generate is going to be built by AI engineers instead of people engineers.

He then added that this will “augment” the people working on it, and, in the future, people will be “freed up” to do “kinda crazy things”. 

Salesforce CEO Marc Benioff also said in December that Salesforce would not be hiring any further software engineers in 2025 amid a “30% productivity boost” from AI tools.

Benioff added that he would be looking to hire more people for sales. 

“If Anything, I Might Be More Valuable Now”

Oliver Fletcher, a developer at agentic AI development company EmergenceAI, thinks Marc Benioff and Mark Zuckerberg’s comments should be taken “with a pinch of salt”. 

“I use LLMs and AI pretty much daily. I’ve seen a big productivity boost. [There’s] absolutely no doubt it is 100% going to change the industry in my view. However, I’m yet to be convinced it’s more than a form of automation that has been going on for a hundred years plus.” 

Oliver said that we have been automating how we work for a long time, and work has changed as a result of that. A few decades ago, people wrote Assembly and then they wrote C and C++ and higher level languages, and AI is now allowing developers to then efficiently bring together bits of code under the guidance of somebody a little bit more experienced, he added.

“I think the key point is that junior and mid-level [employees] are going to really struggle to find work because there’ll be fewer jobs for them.

“I’m fortunate with 10 plus years of experience that I don’t really see myself as junior or mid-level and I see myself becoming, if anything, more valuable. I’ll be of a smaller pool of people that can utilize an LLM efficiently for a great productivity boost.”

Oliver said that he is not personally “too concerned”, and artificial general intelligence (AGI) is all a lot further away than people might think.

“I think that what we have hit is a very efficient form of automation for writing simple code and really accelerating people with experience and that is going to have an impact on junior and possibly mid-level devs… I think I have just enough experience that if anything, this wave might make me more valuable.” 

But he added that junior and mid-level devs are going to struggle, and there may be a point in the future where if we don’t train enough junior and mid-level devs, who (or what) replaces experienced developers like him in 10-15 years? 

“I’ve Made a Career as a Developer on a System That’s Getting Rid of Developers”

Paul Battisson – Founder and CEO of Groundwork Apps – said when he started working with Salesforce around 15 years ago, he showed it to some other developers and they said they didn’t want to use the product because “it’s just getting rid of the developer”.

“I’ve made a career as a developer on a system that’s getting rid of developers apparently – It’s a never-ending thing of developers going to be replaced by something.” 

He said that what Marc Benioff and Mark Zuckerberg are getting at is that AI can help you be a more productive developer if you have it configured and set up properly. 

“I think to say that it could write the level of code of a mid-level engineer is very dependent upon what you ask it to do,” Paul said. “If you give it very clear instructions to do a very specific task, it will write good code for it. But if you say to it, ‘we’re having this sort of problem, how do I go about solving it?’ Unless it’s a problem it has come across before, it’s unlikely to really be able to give you much on that.”

Paul said that this was because these models don’t have the capacity really for creative thought, and they are generative rather than creative. 

He compared using AI tools as a crutch somewhat akin to a student using Google Translate to do their foreign language homework. 

“A: Parts of it would be wrong, but B: The real takeaway there was that you would write it out because you didn’t know enough French or German or whatever to correct it, you were none the wiser, and that’s the same thing here. The code that it generates is only as good as the reviewer who reviews it. 

“It can write decent code, but it can write decent code that doesn’t do anything like what you want. As such, you need to understand what it’s doing for you to be able to say whether it’s good or not. It’s a catch-22 situation.”

What About the Future? 

Even if it’s a tempting offer for tech CEOs to replace lower and mid-level jobs with AI, what might that mean for the future? 

Matthew Jones, an experienced developer who works in the finance industry, told Salesforce Ben that getting into the industry will be “increasingly difficult” if companies start replacing lower-level positions with AI. 

He said: “The companies say ‘We don’t want to hire juniors, we only want these high-end mid to seniors’, but then the problem with that is in 5-10 years time, there will be none of the experience at the lower end. Getting the kind of experience that you need to be able to oversee an AI will be difficult. 

“I think they’re making a problem for themselves in the future.” 

Matthew said that he is glad he learned software engineering at a time when these AI tools were not available, forcing him to learn things he did not know. 

He added: “If you’re just given the answer, you never learn the best way of doing it, and you just take what the AI says and think, ‘Yep, that’s great’.

“I think there might be a problem in the future with this kind of idea. I get that investors are saying ‘Great, I can cut down on all these expensive software engineers because we don’t need them’, but in the future, the pool will be smaller.”

Regarding what Marc Benioff said recently about Salesforce not hiring any more software engineers amid a supposed 30% productivity boost from AI, Matt said: “I wouldn’t want to be a software engineer there where you’re expected to take on more and have these AI agents to kind of back you up. I don’t know, I wouldn’t want to be in that position.” 

But he added that in a straightforward competition between a junior engineer and an AI, the AI will “probably produce better code”. 

“It does come out with a lot of garbage,” Matt said. “Still, it’s a lot better than it was when I first started using ChatGPT. It would make up libraries that look really good and then it would just be like they didn’t exist. But now it’s getting better.”

He said that in three years’ time, the AI models might be much better, and when that time comes, he’ll change his mind. 

“At the moment, I still feel like you need that person. You still need software developers.”

“I Don’t See a Big Explosion of Open Source Developer Tools”

Peter Chittum, Developer Relations Advisor at Westview 1, told Salesforce Ben that in his experience, dev tools often do not get a huge amount of investment, and “certainly” you can say that they are not the biggest money-maker in software. 

He added: “The biggest money makers are the ones that create productivity for big businesses. Developer tools help make that happen, right? But it’s still a step removed from where that actual revenue is coming from. Who’s going to dump the money into creating this super developer productivity thing?

“Let’s say Zuckerberg is right and Meta figures out how to do this for their developers. They all of a sudden have a massive competitive edge and even though they talk a good game around open-source, you’d have to think that they would guard that secret closely to keep on getting as much of an edge as possible against their competitors.

“So maybe all of a sudden, Facebook starts doing it and it slowly sort of propagates throughout the industry. But I don’t see a big explosion of open-source developer productivity tools that suddenly just washes over the market because I think the first company who finds that out is going to be guarding it.”

Peter added that people become experts through experience, and someone does not get to expertise without doing the work day in and day out. 

Final Thoughts 

There appears to be a consensus among developers that AI tools, if they are used to “replace” software engineers in any capacity, will be used to perform the work that entry-level people would do – but will have to be overseen by more experienced people. 

Proponents of AI say these tools will “augment” developers, and it seems this is already happening,

But, ultimately, whatever code an AI – or entry-level developer – produces is only as good as the quality-checker who oversees it, as both are certainly capable of producing a low-quality product.