Artificial Intelligence, the Philosophy of Programming, and the Future of Devs

This is a video version of the content presented in this article that I published on my YouTube channel (in Portuguese).

It’s worth watching! 😁

Between 2020 and 2022, I dove deep into books, especially the ones that discussed Artificial Intelligence (AI). Among the ones that stuck with me the most were Homo Deus and 21 Lessons for the 21st Century, both by Yuval Harari, who also wrote the acclaimed Sapiens.

What’s most curious is that even back then, these books already treated AI as something present, even if with a futuristic tone.

They warned us, still from a bit of a distance, about how AI would transform society. And now, a little over five years later, the topic has stopped being a trend and become part of everyday life.

📚 Recommended books

• Nexus - Yuval Harari
• Homo Deus - Yuval Harari
• 21 Lessons for the 21st Century - Yuval Harari
• The Coming Wave - Mustafa Suleyman and Michael Bhaskar
• The Fourth Industrial Revolution - Klaus Schwab

AI left fiction and entered real life

As a proud child of the 90s, I grew up immersed in movies and shows that portrayed artificial intelligences, usually through a dystopian lens: machines that rebelled and threatened humanity.

That imagery was shaped by decades of cultural content like that. Today, however, we live in a reality where AI is far from destroying the world but is more and more present in our daily tasks.

What used to be fiction is now habit, and that makes me wonder:

How will this technology impact us in the short and long term?

A revolution that’s already among us

The popularization of AI is what impresses me the most.

It’s not a technology restricted to programmers, scientists or big corporations. On the contrary, regular people, of all ages and from all kinds of professions, are using these tools day to day.

Seeing my parents use AI chatbots to do research or learn something new is as impressive as it is symbolic. After all, we’re not talking about a Google that has existed for 20 years, we’re talking about a technology that went global in less than three years.

And I myself, besides the obvious professional use, have been integrating AI into several parts of my routine:

• Organizing my diet and workouts
• Support in content production
• Trip planning
• Managing investments

All of it in a practical, conversational and extremely efficient way.

AI as a personal trainer

A particularly interesting way I’ve been using AI is to track my physical progress.

Every 15 days, I send ChatGPT standardized photos of my body, measurements taken with a tape measure, scale data and body fat measurements taken with a skinfold caliper. From these inputs, both visual and numerical, the model can:

• Identify areas with muscle mass gain
• Detect imbalances or weak points
• Suggest adjustments to calorie intake

Here’s an example of the output from my physical assessment prompt:

Avaliação física – 05/04/2025

Idade: 35 anos
Peso: 86,6 kg
Altura: 1,72 m
% de gordura estimada (pelas dobras e imagem): 10–11%
Massa muscular estimada: ~77 kg
---

Pontos fortes:

- Costas com boa largura, principalmente na região do trapézio inferior e dorsal inferior.
- Braços bem desenvolvidos, tríceps com bom volume lateral, e o braquial aparece bem de lado.
- Pernas com excelente densidade e separação, principalmente quadríceps e gastrocnêmio (panturrilha), já estão num nível avançado.
- Peitoral superior e deltoide frontal bem destacados de perfil.

---

Pontos a desenvolver:

- Densidade na parte baixa do dorsal: a inserção está boa, mas pode crescer mais volume no centro das costas.
- Lats (lateral do dorsal): poderiam abrir mais de frente para criar aquele visual em V mais acentuado.
- Peitoral médio e inferior: visualmente atrás do superior (comum em quem prioriza supino inclinado).
- Bíceps: apesar de bons, têm potencial para mais pico e espessura de frente.
- Ombros: estão fortes, mas podem ficar ainda mais tridimensionais com foco em lateral/posterior.

---

Recomendações:

1. Seguir com o plano alimentar de 3100 kcal por mais 1–2 semanas. Seu corpo está bem responsivo, e o percentual de gordura segue controlado.
2. Priorização de dorsais, braços e peitoral médio no treino. O split atual já contempla isso muito bem, só continuar como está.
3. Pose e postura: quando for tirar próximas fotos, puxe mais o abdômen (sem sugar) para evidenciar o shape e manter a comparação consistente.
4. Foto com contração total (como um duplo bíceps de frente e costas) seria interessante a cada 4 semanas, para avaliar densidade e preenchimento muscular.

---

Se seguir nesse ritmo com os dados, fotos e ajustes que vamos fazendo, vai conseguir ganhar massa magra de forma bem limpa. Pode continuar mandando fotos e medidas a cada 15 dias que vou acompanhando contigo.

In this assessment, the AI identified that my mid-lower chest was lagging behind the upper chest. And this was without me telling it my training split. It deduced that just from the images and muscle patterns, which is insanely impressive.

But here’s a key point: this only works because I’m an advanced user. I know how to build efficient prompts, give the data good context, and guide the AI to generate useful answers.

A beginner might not get the same results.

The hype around creating apps with AI

Today it’s common to see posts on LinkedIn or even Instagram from people with no technical background saying they built an app in 10 minutes with the help of AI.

This is creating a new imagery around building software, something that used to be reserved for experienced devs and now seems to be within everyone’s reach.

And, honestly, I see that as a positive thing: the more ideas get tested quickly in the real world, the better our chances of finding good solutions that will pass the test of time and attract investment.

However, there’s a side effect: the illusion that developing software is simple.

Philosophy of programming: what is developing software?

Before talking about writing code, we need to talk about communication.

Programming is, essentially, communicating abstract ideas clearly enough for a computer (or another human) to understand.

That requires structure, logic, and mental clarity.

Developing software is taking a complex thought, with rules, conditions, architecture, and turning it into something concrete: working software.

Programming is, therefore, a philosophical activity before it is a technical one.

Code is just the tip of the iceberg. The real challenge is in abstraction and in organizing ideas.

Coding was never the hard part

When I started programming 15 years ago, it was common in some companies to adopt the famous “Go Horse” style (Brazilian slang for cowboy coding): writing code as fast as possible to get something live, even if that compromised the quality or scalability of the application.

The code showed up fast, but it was almost never sustainable.

Today, with no-code and vibe coding tools, we can build apps even faster, but the logic behind the creation is still the same. We just swapped machine language for natural language, but the cognitive load remains.

You still have to think about:

• How to store data?
• What’s the user flow?
• What’s the ideal architecture?
• How to scale this application?

No-code isn’t magic, it’s just a new interface

Prompt engineering is nothing more than applying programming logic in natural language. When you write a detailed prompt, you’re structuring your idea the way you would in traditional code.

That’s why no-code and vibe coding aren’t magic shortcuts. They only change how we write instructions, but they don’t change the need for clear thinking, technical knowledge and an architectural vision.

And so we’re back to the root of software creation:

Programming is first and foremost thinking, code is just the result of that thinking

If you made it this far, you may have noticed that this text isn’t just about AI, it’s about how we think about technology, how we communicate with it, and how that shapes the future.

And that future is arriving faster and faster.