When Wikipedia was launched, it changed the game for anyone who wanted to explore the world online, in real time and for free… An incredible treasure trove of information, with new content added daily by an incredible community of people around the world.
Everyone was excited except the teachers...
I remember the skepticism teachers had towards both Google and Wikipedia. They insisted that online resources would never replace “official encyclopedias” (those bound in books) and that a community based on non-professional contributors could not produce content as good and as accurate.
They were wrong.
Wikipedia has emerged as a masterpiece, showing how a community can work together to create the most comprehensive, up-to-date, and freely accessible encyclopedia in the world.
This attitude of my teachers makes me think of today's AI haters in the era of LLMs (Large Language Models). LLMs have already taken over the world and have irrevocably changed various fields: writing, translation, programming, content creation, research, and more.
However, there is a big difference between the Wikipedia revolution of 20 years ago and the current one.
The big difference is simple but crucial: cognitive bias.
Wikipedia was meticulously built over the years by a strong community of experts (and non-experts) who worked together to create a solid knowledge base. If a mistake was made, it was corrected almost immediately, and those who caused problems were quickly banned.
For the Wikipedia community, nothing was more important than facts and free knowledge.
For LLMs, the situation is very different. LLMs are trained from every source of information. Not just real or reliable information, but also information that contains highly biased (and false) information. This is often referred to as data bias or algorithmic bias, as models inherit and reinforce bad patterns.
Also, an LLM will answer even a complex question that it does not know. Its basic architecture is based on probabilistic predictions. An LLM will always have an answer, but sometimes it may contain misleading information.
Just as information found on Wikipedia should always be read with caution (even today), information coming from LLMs should be taken with great caution.
So how do we navigate learning in this new era?
I think we should define basic principles:
Don't just read. Engage with the information. Don't passively consume it.
Think, don't just absorb. Process the information, question it, and connect everything to your existing knowledge.
Develop a critical perspective. Don't blindly accept information. You should evaluate the source, context, and possible biases.
Information acquisition is not a simple matter. It represents the core competencies for growing alongside LLMs.
Whether we like it or not, we are undoubtedly in a new revolution of information canning. So we cannot blindly trust LLMs, and that does not mean we should stop learning.
LLMs are powerful tools and like all powerful tools, they should be used with caution.
LLMs should be companions, not final solutions.
Although the press releases will range from very select to rare, I said I'd pass...because sometimes the editors hide.
