When I was six years old, my mom took me to the Palace of Culture and Science, where I saw a computer :-)
I was astonished to see that this device could handle the simplest logic—it asked questions and checked if my answers were correct, and if they were wrong, it scolded me with a suitable message. It was a revelation, an epiphany that sparked a great passion.
Several Decades Later
I experienced a similar moment of wonder when I gained access to OpenAI’s mechanisms in 2023. Although I had been familiar with techniques for training language models and using them for simple tasks for several years, the scale of the project led to an epiphany that drove further discoveries. Starting to work with large language models, I found that in a short time, I could access a “team” of quickly created bots that communicated and checked each other according to their programmed tasks. This realization allowed for the use of technology in many ways that were previously costly and required substantial human effort.
The Perception of the AI Revolution
To many outside the analytics and data business, it may seem that the "AI revolution" began in 2023. If we asked Google or Netflix, they would be quite surprised. AI has been present in big tech products for many years. The magical moments of accurate recommendations or personalization were already present in Netflix and Amazon products many years ago. Even in more traditional companies, predictive models have been used for a long time.
The Reality: UX Revolution
What we are currently witnessing is a UX revolution. Capabilities that were previously available only to a narrow group of specialists are now opening up to the masses, including latecomers, investors, and others who wish to board the departing train and dream of leaving the competition at the station. Opportunities for a select group of insiders are becoming even greater. How can one secure a spot on this train? One must intelligently leverage the latest advancements in the revolution and apply them to their own data.
The Rise of Automation
This leads us to a startling discovery—models are replacing employees. And that’s a good thing; humans are meant for higher purposes than repetitive work until 5:00 PM.
Experiments with LLMs
A year ago, right after OpenAI introduced ChatGPT, we received a grant for virtually unlimited use of LLM models from this provider. We played like children who had access to next-generation building blocks that no other child had. The results of this long-term research were completely unpredictable, but many interesting solutions emerged, which I will attempt to write about.
Author: Mariusz Jażdżyk
