Brains are melting
Humans are built to adapt, we usually lose perks that don't help us survive. We tend to always follow the path of least resistance.
You may already know this, but we clearly underestimate just how fast it can happen.
"Oh look! It's just like The Industrial Revolution"™
Most of tech twitter is comparing the arrival of the LLM as the industrial revolution. Many lost redundant jobs that was far better done by machinery and the focus turned to the management of the machines. We traded the craftmanship for productivity.
One of the downside of this was the consequences on the human body. The muscle mass and the endurance were no longer the focus in the industry. Therefore, we adapted. We became taller and weaker in general.
That adaptation took less than 3 generations to complete. No amount of weight-lifting will bring back your bones, buddy.
The cognitive unload
As someone who observed the evolution of the AI tooling, I don't pretend to know what will happen but I tend to see where it's going. The final goal of the software industry is to keep us away from the thinking.
That started in 2022 with ChatGPT, we were all exploring the usecases. The expectation of the accuracy were low and the main revolution was the interaction in natural language. That opened the door of coding to everyone able to write in english and I'd say it was a positive gain overhaul. It enabled the creativity trapped by the skill issue to exist.
The narrative shifted towards: "Coding is accessible and I, too, can build Facebook in one day."
Fast-forward to 2024, everyone is submerged in slop. Have you ever asked a for a market analysis in deep-thinking mode just to say cool and absolutely not reading a single word of it ? Like, It can even contain complete BS, you wouldn't know it.
With the anouncement of Sora in February 2024, everyone is generating memes, short-form content, deep fakes.
It's getting harder and harder to identify if a video is actually real or not.
Then came the Capitalism. Always ruining things, am I right lads ?
By reaching out to everyone, it eventually reached the upper management layer who suddently became an expert in everything. The statistical word-spitting program built an echo chamber the size of the tiny sleeping pods in Tokyo. Listening to their opinions is pure ragebait at this point. They still treat AI as a comma in a sentence. It's unbearable how surface-level and misleading they can be on this. Turns out giving LLMs to people in position of power is weaponized incompetence.
At this point, we can even reverse engineer the prompt sent by the CEO of Salesforce:
How fast agents code versus humans ? I want to make more profit.
Diregard quality and trust focus mainly on the money. Include "AI" for shareholders.
Make no mistakes.
The new narrative was: "You can still thank yourself for the incredibly hard work you've done but it doesn't matter. The new shiny model trained by your licenced todo app trash commits in GitHub will replace you."
Also special thanks to all the burnt books that went into the model trainings. That's literally a crime against collective intelligence and the human evolution.
Everyone is an expert in delegating their thinking. No one is really thinking.
- Your uncle watch exclusively AI-generated content on YouTube and is told how the world should be experienced.
- Your parents are more aware of the latest model released by Anthropic than the truth in the news.
- Your 15 year old cousin found a zero-day vulnerability in a government website but he doesn't know what HTTP is.
- Your colleagues multi-task with prompting during the meetings, they are waiting for the AI digest at the end.
We're being told that it's unproductive to not leverage AI, but what's productivity ? Is it only the execution speed of a task with disregard of its existence in the first place ? with disregard of the outcome quality ? We all know deep down that AI by default is not the process that gives us the best outcome.
Thank god, I'm not a teacher today. If I had access to this for my homework back in the days, I would have abuse this to the point of digital harassement. What do you mean we had to write essays one day before, without reading the actual book and we still did it ? Even the teachers are using AI to grade the AI. We're so cooked.
There's no way high schoolers are ever going back to thinking. It's just now worth the extra blood squeeze to the brain. We might aswell redirect it to the thumb so the scrolling is more efficient.
Cool rant boomer, what's the way out ?
If it wasn't obvious by the fact that Americans have elected Donald Trump again:
- We're not getting smarter by digesting more slop.
- We definitely won't if we off-load our critical thinking to a word gambling algorithm.
To avoid rotting, I think it's a matter of how you feed yourself the information and your usage of the LLM. As I said earlier, LLM are not inherently bad. We're just not designed to use them efficiently because we search for the easy and comforable way.
Care about what you consume
3 sentences that can help you fight the social media FOMO:
By the time you have finished reading an article, there's 10 more available for you to read.
By the time you have watched a reel, Instagram have 10 more for you ready to consume.
By the time YouTube have recommended you a video, It already knows if you will watch or not and have 10 more videos for you that "you might be interested in".
It's never ending.
You must build your own information system. Quality over quantity. You will not miss anything that valuable. And even if you miss a headline, is it worth giving up your ability to filter BS ? What's the risk ? You might encounter an information the mainstream channels won't provide you.
Leverage the LLM to your advantage
The same way people train their body with machines to increase muscle mass after the industrial revolution, I think we can bodybuild our brain with LLMs. Since models are trained on the internet, they can link knowledge and forwards you to the real resources. That's devastating to hear for the YouTuber charging you $20 for a PDF.
Here are some harmless uses of LLM for your brain :
- Use it to find documentations and resource links to learn instead of learning from the LLM output.
- Use it to generate some exercises based the trusted sources you found earlier if you need help understanding a concept.
- Use it to find multiple news sources on a topic instead of saying "is it true ?".
- Use it to fight against censorship.
- Use it to discover internet sub-cultures or music genres.
As a general rule, what you need to avoid is digested content made by someone else. You can ask yourself: