My 30-Year Journey Through the Evolution of Artificial Intelligence

Looking back, I realize my relationship with artificial intelligence has been like a long, winding conversation that started in the 1990s when most people thought AI was pure science fiction. But for those of us tinkering with computers, AI had already been whispering to us through LISP functions, Prolog rules, finite automata, and ELIZA’s surprisingly human-like responses. Do you remember joking around with the Doctor program loaded up through GNU Emacs, amazed that a few lines of code could almost feel like talking to a therapist?. All of those weren’t just tools or programs. They were milestones in a time when programming felt more like a philosophical dialogue than a technical task.

What began as curiosity, playing with these early AI experiments, has evolved into a deep passion that continues to drive me today. I owe much of this journey to an incredible professor, Patrick Greussay, who taught me to find beauty in extravagant concepts, elegance in algorithms, and see the recursivity hidden in Bach’s “Well-Tempered Clavier.” He showed me that AI isn’t just about code, it’s about recognizing patterns, finding harmony in complexity, and understanding that the most profound innovations often come from connecting seemingly unrelated ideas.

The Wild West Days: Building Something from Nothing (1990s)

In 1994, while working on my master’s degree, I had this crazy idea to create a voice programming language called “ProgSpeak pour la programmation parlée.” #INRIA was kind enough to lend me a voice recognition hardware, this massive 1kg beast that picked up every sound imaginable. I ended up programming a small Meta Language for basic math operations using Sumerian language structures (don’t ask me why, somehow it just clicked with C function prototyping!).

The funny thing? I had to sneak into the lab after 2 AM because that was the only time it was quiet enough for the system to work. But honestly, those late nights were magical. There’s something incredible about building a programming language from scratch, diving deep into linguistics, and exploring wild ideas about how humans and computers should communicate.

A year later, for my postgraduate thesis, I got even more ambitious. I worked with a group of future IT engineers, using ethnomethodological approaches to figure out the best way people should talk to computers for development work. The conclusion? We needed something more universal, maybe Esperanto, Volapük, or even just plain English. Great idea, but I was about twenty years too early! 😊

The Reality Check: Learning to Build Real Solutions (1995–2014)

Then the IT business world called, and I answered. For nearly two decades, I learned what it really means to deliver solutions that work in the real world. You learn to enhance products, live in the present moment, and guide teams toward projects that actually matter to people. It wasn’t glamorous research, but it taught me something invaluable: the best technology is useless if it doesn’t solve real problems.

Coming Home: The Deep Learning Revolution (2014–2023)

By 2014, I felt that old scientific itch again. I went back to the same lab, working in Natural Language Processing — familiar territory, but with exciting new possibilities. Thanks to #NVIDIA and gaming laptops, I could finally run decent ML algorithms without needing a supercomputer.

I completed my PhD on sentiment analysis, multilingual and multi-topic opinion classification using deep learning, word2vec, and embeddings. Then everything exploded: BERT, Transformers, LLMs!

I spent years teaching AI because I felt I had to give back somehow. Covering most of the subjects since the 1950s, perceptrons to 1986’s backpropagation, through CNNs, LSTMs, GANs, I wanted to share the excitement of this incredible journey.

The Pattern I’ve Learned to Recognize

Here’s something fascinating I’ve observed: AI moved in these long cycles. The first AI era took 30 years, then backpropagation gave us hope in the 80s, but we had to wait nearly 30 more years until 2010 for the deep learning breakthrough. We got a solid decade (2010–2020) to experiment, test, and validate our ideas.

The Explosion: Why I Had to Jump Back In (2023-Present)

Then 2020 happened. LLMs went public (the API), and suddenly we’re in this crazy AI gold rush. Multimodal models like GPT-4o, Sora, Claude 3 Opus, Gemini 1.5, they’re arriving faster than I can keep up with. Universities can’t compete with Big Tech labs anymore, and that’s when I realized it was time to jump back into the business world.

Right now, we’re seeing this exponential explosion of AI tools (mostly wrappers around the big models, but still). And honestly? I think we’re maybe a year away from #AGI and #ASI changing everything we know about work, creativity, and human potential.

What This Journey Has Taught Me

After three decades in this space, I’ve learned that the most exciting AI projects happen when you combine deep technical understanding with genuine empathy for real-world problems. I’ve seen AI from the research lab to the boardroom, from those 2 AM voice recognition experiments to helping companies transform their entire business models.

But here’s what keeps me up at night (and not just because I’m testing voice recognition systems): we’re approaching a crossroads that feels straight out of science fiction. Remember HAL 9000? That calm, rational AI that knew exactly what needed to be done? We’re getting closer to that level of AI capability, but the big question isn’t whether we can build it, it’s whether we can control what comes next.

Will we find ourselves in a world where AI systems make decisions we can’t understand or override? Or will we figure out how to harness this incredible power while keeping humans firmly in the driver’s seat? Having watched AI evolve from those clunky 1kg voice recognition boxes to today’s sophisticated language models, I’m both thrilled and sobered by how fast we’re moving.

The best part? We still have time to shape this future. Every conversation I have with business leaders, every project I work on, feels like I’m drawing from this incredible three-decade foundation while helping write the next chapter of human-AI collaboration.

That’s what keeps me passionate about this field, we’re not just building technology, we’re defining the future relationship between humans and artificial intelligence.