World is Changing. Software is Changing.
This is crazy isn't it? how it has changed quite rapidly in just a few years? We literally went from .0 to version 2.
To clarify before we start, 1.0 is a must for 2.0 and 2.0 is a must for 3.0 and this is not going to change for a long time (and shouldn't). It is not like a product versioning, but rather an additional layer in a technology stack. Wheel + motor + aerodynamics.
This means we have made the fundamental change (and it is not yet complete, but fast growing, as our end goal is software 3.0).
This will require much more expertise as the problems we'll be dealing with will be greater.
It is not enough to just know how to write code in a programming language anymore. That was software 1.0.
Humanity has come a long way since the first programmable computers. The other day I found this old Toshiba computer at the office.
It turns out as a kid, I was the one using it. I had no memory of it, but as soon as it turned on the computer and Windows Vista came up, then there was this user with a penguin image. I instantly knew it was me. I was 8-9 years old probably.
That was software 1.0, This platform is software 1.0 too.
ChatGPT, Grok, Gemini, DeepSeek user webapps are software 1.0 wrapping and running a great example of 2.0.
Next move is gonna be software 3.0:
Structured prompt -> program -> mission completed.
We haven't yet figured out how to achieve 3.0 directly but we have a temporary solution where we use 2.0 to generate 1.0 to simulate 3.0.
I repeat, we use 2.0, to generate 1.0, to simulate 3.0.
To make these more concrete, I will give direct examples:
I can integrate GPT 4o (one of the ChatGPT models) into an IDE (where you write code) and I can use GPT and systematically ask it to complete parts of a software by writing the code section by section, back and forth, simulating a development phase.
So far, we used 2.0 (GPT) to generate 1.0 (direct program source code in a programming language).
Now we are at the last step, simulating 3.0.
This is actually the end result of steps one and two.
When the output program runs smoothly, this means we simulated a software 3.0 workflow. But it isn't yet 3.0.
Why?
Because to be fully doing software 3.0, I shouldn't be using an LLM to generate computer code, but rather be using an AI tool (which may or may not include an LLM inside) that UNDOUBTEDLY outputs PERFECT programs.
This means "hey, I need an app that tweets hello to all my followers every sunday" and the computer responds with "done. visit it as hello-tweeter.com".
And you just literally visit the link and everything is working.
No code, no failed attempts, no debugging.
So, software 3.0 is magic and we will no longer need to know anything about anything?
Not exactly.
With software 3.0, you will be following these steps to be remarkable and beneficial to humanity:
1- train your brain with knowledge, problems, live life and experience things.
2- focus on topics you like (such as marine biology, astronomy, literature, how computer memory works, how an LLM works, how an airplane flies better than other airplanes, how an algorithm design works for a programming language, how manufacturing robots work, and so on).
3- spend a very long time on these (years) by studying them, innovating, trying, failing.
4- think of all the possible fixes and enhancements you came up with over these years and your trials.
5- make the ultimate enhancement to these fields by applying your innovations/fixes/enhancements.
6- let the next generations continue this loop.
So... these sound like... general life goals? What does it even remotely have to do anything with software 3.0 or software overall?
Because software rules the world. Software will be in every single aspect of your life. It is 90% there already.
Everything you do has a software aspect in the modern era. It will become 100%.
You go to doctor, he uses software to do his job.