In a leaked recording, Amazon cloud chief tells employees that most developers could stop coding soon as AI takes over
Matt Garman sees a shift in software development as AI automates coding, telling staff to enhance product-management skills to stay competitive.
Read Full Article...C4AIL Commentary
Sound bites made good headlines but rarely inform well.
- Code Generation Capabilities are indeed rapidly maturing and are injecting massive productivity gains into the profession.
It’s a fairly conservative statement to make that software engineers will spend less time focusing on writing code in the future. To see that this is the case, a simple look at the second generation of coding tools such as v0.dev or postgres.new is enough.
Hand constructing web components from code has long been on an automation trajectory, AI is just finishing the job for the 85% usecase.
It is also clear that the potential for productivity gains is not nearly exhausted - and while model quality will certainly improve and drive further gains, we believe that the real impact is coming from increasing comfortable scaffolding and integration with developer tooling, for which much opportunity still exists.
Uneven effect.
That said, by our own experiences, the productivity gains from AI are unevenly distributed, with high skill / senior engineers experiencing a much stronger productivity gain than less experienced ones.
This stands in contrast with lower skill ceiling jobs such as customer service in call centers, where introduction of AI is able to elevate junior employees to intermediate senior performance - thus driving redundancies on the higher end of the pay scale.
AI code generation works - until it does not and the point of failure is not at all easy to discern for a non expert.
Where this highlights a sustainability challenge for the AI movement is that people are not born senior talent and, like in any other industries, tasks currently used to grow the next generation of talent, is falling to the AI + Senior Engineer productivity gain much faster than senior roles.
About Agents
Agents, often brought up in context of code generation, continue to be non viable without human in the loop interaction - the individual reliability of LLMs, foundation for code generation, continues to be off by more than an order of magnitude to support full automation. Until such time that reliable triple-9 performance can be reached, agents are aspirational.
The end of the Engineers as Pokémon era.
People, especially in technical roles, are the second largest cost center for Silicon Valley companies and more flexible than Infrastructure when it comes to driving short term savings.
It should be of no surprise to anyone that aggressive investments are made to reduce people costs after the last two years. The AI revolution in Big Tech has primarily taken the shape of Infrastructure (GPU) investments replacing talent and product capabilities as the main frontline of competition.
Where once rapid response product teams were deployed to quickly match other tech giants moves in the market, leading to the famous hoarding of top talent like rare pokémon cards, building the largest GPU capabilities (and denying the same to the competition) has become the new law of the land.
At the same time, AI has become the hail mary for tech companies whose valuation just 2.5 years ago was beset by end of growth fears and spectacular implosions like Elon Musks hostile takeover of Twitter.
Thus the companies writing the narrative and directing the thrust of the technology are singularly united in pushing forward code generation as a frontier delivering them future bottom line.
Code Generation is the best case for AI
Programming languages have clear structure, small vocabularies and their entire history, best practices, source code and application are public.
Github, Stack Overflow and the public web hold more high quality, annotated and structured training data - in code, issues, kernel mailing lists, Q&A and documentation than any other field.
Much of recent research tells us that quality of training data is crucial for model performance, often outweighing scale.
As such it should come at no surprise that software engineering is one of the most exposed fields when it comes to AI disruption.
Much more fundamental changes on the horizon.
We’ve written about how software itself is about to change through the technology a few times and it’s worthwhile to revisit: Fundamental changes due to the democratization of General Purpose Compute are likely to have a much bigger effect on the profession down the road than short term efficiency push.
The challenge for companies
Exhortation of the impending demise of software engineers is as misguided as saying “gunpowder made soldiers obsolete”.
However, the constant drumbeat of “Your job is doomed” is having a clearly visible negative effect on employee engagement - and they have good reasons to
Companies are well advised to enable their own employees to engage in upskilling - the illusion that AI talent will be available in the market to replace existing employees is a very expensive pipe-dream ignoring the harsh reality of a completely overwhelmed education system unable to keep up with a rapidly changing technology.