Generative AI is supposedly coming for our jobs. But, if Silicon Valley bosses are to be believed, software developers may be especially at risk of redundancy.
Late last year, Marc Benioff, the Salesforce chief executive, said his company would not hire any software engineers in 2025 thanks to advances in AI. Meanwhile, the CEO of Nvidia, Jensen Huang, suggested there’s little point in teaching children to code, because AI will soon be capable of performing the majority of coding tasks.
What is vibe coding?
‘Vibe coding’ is an early example of how AI systems might come to dominate the world of coding. Tools such as Cursor are enabling users to create code through prompts. Would-be developers simply describe the kind of software they want to create and the platform does the rest – it even fixes bugs.
According to the former AI director at Tesla, Andrej Karpathy, who coined the term, vibe coding is “where you fully give in to the vibes; embrace exponentials and forget that the code even exists.
“It’s not really coding,” he added. “I just see stuff, say stuff, run stuff and copy and paste stuff and it mostly works.”
However, Karpathy cautioned that the vibe coding methodology is better suited for “weekend projects” than it is for more serious endeavours.
Vibe-coding challenges
Anyone with a simple understanding of software development can give vibe coding a go by playing with tools such as Cursor, Replit or GitHub’s Copilot. Unlike the coding days of yore, when developers had to study intensely to gain a working knowledge of their language of choice, beginner vibe coders can create an application that works well enough in just an hour or two.
As it enables people with only a basic understanding of coding to spin up new projects through descriptions and prompts alone, vibe coding marks a fundamental shift in the barrier to entry for simple coding.
However, while this AI-assisted approach might accelerate otherwise laborious low-skill coding projects, it’s unlikely that platforms such as Cursor will eliminate the need for knowledgeable software engineers.
Vibe coding appears to deliver on the Silicon Valley promise that AI is best suited for low-stakes high-labour tasks, with coders working in tandem with the algorithms to refine and improve code bases.
Sabrina Farmer, chief technology officer at Gitlab, is confident that software engineers are safe in their jobs. Skilled software engineers are still needed to oversee and determine the strategic direction of their projects. Plus, the code base generated by machines may be far from optimal, or hiding bugs or security holes that only a seasoned pro could easily spot.
Moreover, sentiment around AI seems to be evenly split among software engineers, according to a recent poll from Wired. Although the vast majority of those surveyed said they use AI once per week, 38.4% described themselves as AI pessimists and 26% as AI agnostics. But AI optimists such as Farmer argue that coders are happier and more productive when they can delegate the “grunt work” to machines.
Generative AI is supposedly coming for our jobs. But, if Silicon Valley bosses are to be believed, software developers may be especially at risk of redundancy.
Late last year, Marc Benioff, the Salesforce chief executive, said his company would not hire any software engineers in 2025 thanks to advances in AI. Meanwhile, the CEO of Nvidia, Jensen Huang, suggested there’s little point in teaching children to code, because AI will soon be capable of performing the majority of coding tasks.
