Listen to this story
|
The barrier for entry to become a developer is dropping everyday. The most recent phenomenon that everyone is still talking about is Anysphere’s Cursor AI coding tool, which has basically made everyone a developer. Now, there are more tools coming up in the same category such as Codeium, Magic, and Zed AI, all of them trying to come up with the same formulae.
This definitely brings in the question – what would happen to the software developers of today? Graduating out of colleges with computer science degrees in a world to compete with the people who are becoming software engineers with AI tools, the turmoil of an average software engineer is real.
The solution is easier said than done – upskill yourselves and focus on higher order things such as building foundational AI. Even 8-year-olds are building apps using Cursor AI in 45 minutes.
A Profession Like Never Before
Since there is no barrier to entry, no degree requirements, and no regulations about who can join the market, software engineering has become a profession that has never happened ever in the history. There are plenty of opportunities for developers to upskill.
But “upskill to what?” is what people ask.
The conversation from LLMs to SLMs to coding assistants to AI agents keeps changing so swiftly, it can be challenging to determine which new skills are worth acquiring. This question reflects a broader uncertainty about how to prioritise learning in a field where the next big thing seems just around the corner.
Saket Agrawal, a developer from IIT Guwahati, said that it is not as much about the technological shift but the advancement of automation tools that reduce the time and efforts for the same skills. “I don’t see any big threat to existing software skills suddenly and software has been the field all the time which needs continuous skills updation based on requirement without leaving your old skills instantly,” he said.
Another user on X put it in a funny way. “Software engineers need more updates than my grandma’s Windows 95. Ever tried explaining AI to her? It’s like defining gluten-free bread to a caveman!”
It is widely discussed that a lot of software engineering jobs are dying. “Winter is coming for software engineering,” said Debarghya Das from Menlo Ventures, saying that many of the current software engineering jobs would become a distant memory.
Scott Stouffer adds another layer to this conversation by suggesting that some are experiencing an upgrade in their lives at a pace that surpasses others. This notion of being “upgraded” faster could imply a divide between those who adapt quickly to technological advancements and those who struggle to keep up.
LLMs to Upskill?
While there is a very interesting caveat to all of this conversation around upskilling. Hardcore skilled developers believe that leveraging tools such as Cursor and others can take them to another level where the new developers would never be able to reach. Yann LeCun has already told developers getting into the AI field to not work on LLMs.
Andrej Karpathy recently said that the future of coding is ‘tab tab tab’ referring to auto code completion tools such as Cursor. Further in the thread, he added that with the capabilities of LLM shifting so rapidly, it is important for developers to continually adapt the current capabilities.
Some people are sceptical if they even should get into the computer science field anymore. “…if I was new to programming I would be too tempted to skip actual learning in favour of more LLM usage, resulting in many knowledge gaps,” said a user replying to Karpathy. This truly feels like the true way forward for many developers.
This is similar to what Francois Chollet, the creator of Keras, said a few months ago. “There will be more software engineers (the kind that write code, e.g. Python, C or JavaScript code) in five years than there are today.” He added that the estimated number of professional software engineers today is 26 million, which would jump to 30-35 million in five years.
This is because developers who are proficient with coding without code generators can never be replaced. People who built programming languages and foundational tools are still very well versed with coding than people who are just using Cursor to build apps. Sure, there can be an abundance of people building apps in the future, but the scope would just be limited to that.
Meanwhile, highly skilled 10x developers would be focusing on leveraging such tools, or possibly finding flaws in them, to create even better software. So to say, creating the next Cursor or ChatGPT.
There is an abundance of things that can be done. For instance, focusing on enhancing hardware or building infrastructure for running future workloads can only be comprehended by experts in the field. For example, companies such as Pipeshift AI, Groq, Jarvis Labs, and many others who are working on different problems than coding.
The truth is that such AI tools can never replace human intelligence or jobs, only augment them. “Generating working code is only a part of the responsibility,” said VJ’s Insights on a post on X. Though “Yes, if you are someone who *just* writes code, you need to start thinking differently.”
In the near future, there are predictions that the future of software engineering would be about managing a team of AI agent engineers, and telling them how to code. This will make every engineer akin to an engineering manager, delegating basic tasks to coding agents while focusing on higher-level aspects such as understanding requirements, architecting systems, and deciding what to build.
It is high time that software engineers start upskilling themselves, and currently, it looks like using generative AI tools is the best way forward, not without them. Who knows, you might also become a solo-entrepreneur building a billion dollar company alone.