How did Google (unintentionally) build the foundation of ChatGPT? 😯
Machine translation from the original post. Not reviewed by a human translator. Historical claims may no longer be current.
How did Google (unintentionally) build the foundation of ChatGPT? 😯
A bit of history: in 2017, a team of researchers at Google published a research paper with a somewhat boring title: "Attention is All You Need".
What they didn't know then is that this paper would become one of the greatest breakthroughs in the world of artificial intelligence.
In the paper, they presented a novel model called the Transformer, which managed to understand human language far more efficiently than previous models.
The central idea of the transformer model is that instead of reading text linearly, this model used a "Self-Attention" mechanism - a system that lets it understand all the contexts in a sentence simultaneously.
And that changed everything.
The model became the foundation of all the advanced AI systems we know today:
From Google Translate, through smart search systems - to ChatGPT.
By the way, there's an Israeli touch here too:
Among the eight researchers who wrote the paper was Noam Shazeer, an Israeli scientist who was part of the development team at Google.
So yes, Israel was there already at the starting point of the AI revolution.
But the ones who took this idea the furthest were OpenAI - who developed GPT-3, GPT-4 and additional models, using the principles of the transformer.
And of course made it accessible to the general public through a convenient and free (in limited form) interface at the end of 2022. And the rest of the history since then you already know.
So yes, it's pretty amazing to think that the foundation of the technology that powers ChatGPT came out of research by Google itself.
Do you understand the irony?
Google was the first to identify the potential of transformers - but OpenAI was the one that turned them into the world's standard.
And it's not the first time this has happened.
This story is very reminiscent of what happened in the 80s with Xerox, Apple and Microsoft:
💠 Xerox PARC developed the graphical interface (GUI) - including windows, icons, and a mouse - but did nothing with it.
💠 Steve Jobs saw the potential, took the idea - and created the Macintosh.
💠 Bill Gates identified the opportunity, developed Windows, and conquered the market.
The same process happened here:
Google developed the transformers - but did not lead the revolution.
OpenAI took the idea - and created ChatGPT that changed the world.
And now? Google is trying to close the gap with Gemini.
Moral of the story?
A good idea is only the beginning.
The one who executes it right - is the one who leaves a mark.
It's all about execution.
The question that remains open is whether Google can still turn the tables like what happened between Windows and Mac?
Or whether this is going to be yet another failed attempt like the social network Google opened following Facebook and closed in disgrace, with no active users, a few years later.
(Remember Google+?)
I can attest for myself that in all the attempts I've had so far with Gemini I find myself going back to ChatGPT, so Google still has a long way to go to win our hearts. 🤷♂️
