Paradigm Shifts
In our lifetimes we have witnessed the following paradigm shifts in tech:
- The Internet revolution (late ‘90, early ‘2000)
- The Mobile shift (from Desktop) (‘2010)
- The AI wake-up (‘2020)
Which one do you think is more significant? Let’s make an imagination exercise for the long term, let’s say for the next 100 years:
Will the Internet still exist in 100 years? Of course it will but it will be just like electricity, always there, never to be thought about.
Will the Mobile phones still exist in 100 years? The form factor will still be relevant since a mobile display is bigger and can be much more practical than a wearable device, like a Watch for example. But probably it will not exist in it’s current form.
What about AI? AI is just starting, although we may feel it’s already mature, it’s more like a toddler now. What AI changes:
- Intelligence commoditized! Intelligence will less and less differentiate people, AI know-how will.
- Entire categories of jobs will be gone, millions of jobs lost or shifted! Industries most impacted by AI: software, law, accounting, medicine, and all the white collar jobs.
- Learning on steroids. You can learn any new field 10x faster since AI will create a tailor made learning experience for you. You can now easily learn about programming, the stock market, medicine, rocket science and what not.
Nothing else in the world…not all the armies…is so powerful as an idea whose time has come.
– Victor Hugo, The Future of Man
Intelligence Commoditized
White collar jobs at risk. Firstly AI will be a tool for professionals, for the first 5 – 10 years. After ~10 years AI systems will completely take over some categories of white collar jobs! Although new jobs will be created for AI.
Humanoid robots still have a long way to go before they can replace humans. Sure robots are very good at repetitive tasks, like the tasks done in a factory. However dexterity, such as handling tools, and reliability, such as not falling on kids or pets, remain very hard technical challenges. Unlikely to be solved in the next 10 years.
This means that a plumber, constructor, or gardener can be paid more than a programmer, accountant or lawyer because they do hard physical work that will not be replaced by humanoid robots in the next 10 years! This sounds like a communist utopia!
The Software Industry
In product development writing code used to be the bottleneck. I postponed features for years and years because I needed to understand every system inside and out. Programming is notoriously hard and time consuming.
Now AI tools like Claude Opus, OpenAI’s Codex and Google’s Gemeni Pro greatly augment a programmer’s ability to write, refactor and debug code up to the tune of making him/her 10x more productive! Next, in a few years, programmers that only write code manually will not be wanted anymore by companies seeking efficiencies.
It’s sad for me because manual coding used to be the job that I loved more than anything else. Now it’s going away, and fast!
“Coding” was never the source of value, and people shouldn’t get overly attached to it. Problem solving is the core skill. The discipline and precision demanded by traditional programming will remain valuable transferable attributes, but they won’t be a barrier to entry.
– John Carmack (legendary coder behind Doom & Quake)
Soon the bottleneck will shift to AI/Agents know-how, marketing, UI, UX, testing and combining all those disciplines together to make a software product. You can no longer be a simple programmer, just writing code, you need to be multi-disciplined!
It doesn’t mean you shouldn’t learn programming and how computers and AI work, it means that is no longer required in order to make a software product. Deep Blue beat Kasparov almost 30 years ago and people still learn and play chess today.
AI in Software Development
We first started using AI at the end of 2022 with ChatGPT 3.5. We copy & pasted functions from ChatGPT into our own code. We pasted code inside ChatGPT and he gave us ideas. Many programmers still use it like this today.
Then we used Microsoft Copilot inside VS Code to write boiler plate code for us and talk to the AI inside the IDE without switching to a browser. Just a minor improvement.
But today we use Claude Code and Codex inside VS Code, Cursor or from the CLI to manage an entire project. AI can now implement entirely new features or fix bugs across our entire codebases! It’s like magic when it works, and it works most of the time.
It’s not 100% there yet but many times when AI doesn’t work it’s because the human in control doesn’t understand how the agent works and that he has to give it a better context and a better description of the feature/problem that needs to be implemented/solved.
The Future of Coding?
In the last few months since Opus 4.5 was released, the writing is on the wall: AI will solve programming! There’s no question about it… unfortunately…
But will developers lose their jobs? There is a huge need, from companies around the world, for old systems to be upgraded or completely new products to be built. According to Jevons paradox when technological advancements increase the efficiency of a resource, the total consumption of that resource rises rather than falls.
If AI can make developers 10x more efficient there will be 10x more software written, not less programmers! You will now be the manager/orchestrator of multiple Agents doing multiple jobs at the same time, you will see the bigger picture and no longer be coding in the trenches.
So as the R.E.M. song goes: It’s the end of coding as we know it (and I feel fine)!