I feel like the only people telling me that AI is taking over software development are salesmen and people that don't use the tools and aren't experiencing what bullshit they are.
You auto accepted AI GENERATED code, you didn't read the code, you released it to production and you think this was a good idea?
VC's and podcasters are far easier to replace with AI than software engineers.
Now if only it actually worked. Believe me, I'm programmer by trade but I'd love nothing more than to tell an agent what I'd like the experience to be and briefly describe the architecture that I think best supports that and wait a couple of minutes and see that translate into code. But it doesn't work. Instead you spend more time than ever chasing weird bugs you didn't even know was possible to create.
We keep focusing on how AI will destroy software jobs... but why are people completely ignoring the fact it can replace other industries even faster?
Most of the time when I'm in the weeds of "agentic" code or "vibe coding", I find myself wishing I would have just written it myself. Even with very clear code refactorings that any junior dev could do with 100% accuracy, AI will randomly misname something, drop a property, use null instead of undefined (for nodejs), invert an if, etc, etc. It makes SO MANY mistakes that you really have to go through with a fine-toothed comb and by the end, that "10x productivity" really goes back to "1x" or potentially below and much more risky code that you then need to review more closely
'Software is ultimately about improving some human's experience.' This is such a great point and reminder, and is true about every economic activity. I'll add that as long as human experience can be improved, they will pay to have it improved, so we'll have plenty more to do.
If you were 25???? I am 65 and loving it! Really appreciate your suggestions at the end -keep up to date and get used to identifying human problems. Great job guys - thank you
My only counterargument to the agriculture analogy is food production has gone up mainly in the dairy, grains and meats production which can be automated but what I'd consider highly dense nutrients foods mainly vegetables and fruits haven't really scaled. You have plenty of DTC brands selling highly marketed, specialized products as a branding play vs generic products. Software might be similar in its easy to get cheaper options, but many products do well when they're highly specialized.
It’s the first book that made me feel both terrified and unstoppable. Hidden Codex of Wealth by Dorian Caine didn’t just give me knowledge—it gave me clarity. And once you see what’s really going on… nothing ever looks the same again.
Every time I see clueless tech bros talk about anything beyond the scope of code, the more I believe the modern day Dickens novel villains are the CEO's of Tech Start Ups.
Exciting stuff! We are in the vertical of engineering simulation software and most legacy products have tons of ancient code (20-40 years). Pondering how AI will play out in 1-3 years - 1) New products written from scratch will win, 2) Existing code bases will be rewritten 3) Nothing changes - the existing products are too big for new products to compete (even if developed with AI/vibe tools) and any existing code base is a house of cards and cannot be touched. What do you think?
Thank you for sharing. Financial education is crucial today to show incredible resilience and discipline in the volatile market, masterfully balancing strategy and insight for success. This dedication to continuous learning is inspiring...managed to grow a nest egg of around 200k to a decent 632k in the space of a few months... I'm especially grateful to Evelyn Infurna, whose deep expertise and traditional trading acumen have been invaluable in this challenging, ever-evolving financial landscape.
It's really sad to see that rage-bait tweets are becoming accepted in a professional settings. If software engineering deserves the title "engineering", it is obviously not about harvesting the same crops (React apps?) over and over, but the act of creating the harvester itself. The reason why we talk about "AI" replacing software engineers instead of other professions first is that the software engineer's product is simply text in a context-free grammar and so anything that can write something in that grammar will look like it could replace software engineers. People doing anything proper like research in the software engineering field know what truly goes into the profession when not just creating web apps. A fact that people at a company called "Y Combinator" should be well aware. It's also funny to realize that Y Combinator doesn't even use the word "LLM" anymore. Instead, they hype-up next token prediction to automatically lead to AGI in the next five years. Transformers are an impressive innovation step in AI, don't get me wrong, but we had the same LLM architecture for years now and people are slowly realizing its innate flaws. I'm not saying there is not going to be a software-engineer-replacing AI in a few decades, or years, or even months. But it will take a breakthrough and maybe one that will lead to true AGI and then all bets are off anyway.
Oftentimes when people say that a profession will go extinct, they severely underestimate the complexities of said profession. Software engineering is more than just writing React and Ruby code and deploying it whilst crossing your fingers that it will work. If this would be everything they do, we could just call them programmers. Creating websites using AI is now a reality, but creating rocket guidance software, novel algorithms, etc. will not be done by LLMs. Maybe they help auto-complete some parts, but not more.
Looks like outsourcing can no longer needed.
I love that improvements are in a 6 to 12 months scenario not like 3 to 5 years. It’s crazy fast. Exciting but also scary. But I keep studying, understanding, experimenting, doing my best and that is it. 🤷
Crazy how this info is free. I love the internet
I do think most people who argue that these AI tools are just not good, they assume that the tools will never get better? I mean just 2 years ago we didn't even have none of this. And yet the tools have progressed this much. Why would you think in next 5 - 10 will still remain the same?
@chapterme