Man Reflects on Andrej Karpathy's Predictions About How Knowing How to Use AI Will Divide Society: ‘Give it 5 years. How many of those 80% will still be congratulating themselves for being too wise and righteous to use AI?’

Author image for Olivia Arocena
Olivia Arocena • Writer and Editor
Oct 17, 2026 8:00 AM EDT

This entire AI debate has become a battle over Intellectual Identity and style

The AI vs. SI debate has made us question what it means to be human, to write like a human, or to work as a human. Once bots and agents took over our jobs or significantly changed the way we do them, we started wondering what we bring to the table.

Both being technologically enthusiastic and being skeptical can be ways of oversimplifying the issue. Are we using AI, or using it to prove we're smarter than everyone else? Who is right? People who say we need to learn to work and live with it, or people who say we should focus on making our human capabilities more valuable to stand out in a world full of slop?

I personally think both parties are right, but not in the way you think. We should definitely embrace AI as an assistant at work, except when our job is something that doesn't require AI at all. But I'm also one of those people who think that since AI arrived, human-made stuff has become more valuable, and true creativity has proven itself as un-mimicable by a machine.

AI is known for rapidly changing and evolving, so learning everything about it today wouldn't be a little paradoxical? All of that, in two years, will probably be completely obsolete. We should learn to constantly adapt, stay aware of its advances, and also, why not, of its political implications.

What baffles me is that all the people who discuss the future seem to be too confident about what it will look like, when ten years ago we could've never predicted this SI(AI) revolution.

Artificial Intelligence is definitely changing our thoughts about what it means to be smart, and I like the debates it's sparking. I definitely think the people who think they know exactly how AI works or trust everything it says are the ones in danger.

Perhaps I'm a romantic at heart, but I believe in humanity; I believe in its imperfection and its power to produce knowledge. AI can put together a bunch of words and sound coherent, but it can't think or make new associations; it just works on prompts and preestablished dogmas. I don't think human knowledge can be reduced to logical operations.

We've created language, and language is a beautiful thing. It's plurivalent and polysemic; I'm not sure machines will ever be able to understand that. They can understand formulas and relationships between ones and zeros, but not meanings, feelings, or complex relationships between image and discourse.

And I don't mean we're better than AI because we feel things; I literally mean we can find more meaning in language and make many more associations than a machine. We should use it as a tool, an incredible tool that stores data and can quickly look through it, but nothing more.

While I was writing, I thought about Borges and his story 'The Library of Babel'; I think he might have predicted AI.

Advertisement
  • AI Zach Malinowitz
    via Zach M

    How closely we will work with AI

    Stock photography for depiction only

  • By the Time Everyone Catches Up, It Will Already Be Too Late

    According to Andrej Karpathy's recent approximations: 80% have no access to Al, no interest in it, or actively spurn it. 20% casually only see
    via bigdickzillionaire
  • free-tier derpy chatbots as a better Google search or writing aid. Just 0.2% have seen agents complete complex projects with a prompt that used to take weeks or months.
    via bigdickzillionaire
  • And somehow, people who know the least about Al are the most certain about what it can't do. "It's just autocomplete bro." "It can't tell the truth."
    via bigdickzillionaire
  • "Everything it makes is slop." Guess you tried ChatGPT in 2023 and apparently learned everything you'll ever need to know. We've all heard about the
    via bigdickzillionaire
  • technological divide. My grandmother never touched a computer. Anything beyond calls and texts on her smartphone meant another call about how "the d in thing is broken again."
    via bigdickzillionaire
  • But she got along just fine. Worked until retirement. Lived independently until hospice. She could ignore the computer revolution because it unfolded over decades. She was fine. We don't have decades.
    via bigdickzillionaire
  • Give it 5 years. With recursive self- improvement, even that may be generous. How many of that 80% will still be congratulating themselves for being too wise and righteous to use
    via bigdickzillionaire
  • AI? Sure, someone starting in 2028 will have access to vastly more capable systems than we have today.
    via bigdickzillionaire
  • 00
    via Mushvig Niftaliyev

    Workers learning about AI

    Recreated scenario included to enhance reader understanding

  • But access to the technology isn't access to the opportunities it created. Markets captured. Industries restructured. Economic advantages consolidated. None of that
    via bigdickzillionaire
  • resets when you finally catch up. And being an Al power user doesn't guarantee you're among the winners either. ,,But Al will be easier to use by then!"
    via bigdickzillionaire
  • Yes. Obviously. But nobody is promising the opportunities would still be there when you finally showed up.
    via bigdickzillionaire
  • We're watching the foundations of a two- tier society being laid while most people haven't even realized there's a race.
    via bigdickzillionaire
  • My grandmother could afford to ignore a technological revolution that took place over 50 years You're betting your future. on having the same luxury
    via bigdickzillionaire
  • amp1212 If you believe in the "strong singularity" -- eg an RSI powered leap to superintelligence, with its increases in capability increasing by the day... then what you or I know of Al today won't matter.
    via bigdickzillionaire
  • The poet who writes in longhand and never uses a computer and Andrej Karpathy would be similarly placed with respect to post-singularity Al. If you actually believe in a "strong singularity" -- then the time you've spent learning all sorts of skills in the "pre- singularity" era are akin to,
    via bigdickzillionaire
  • say, someone who knows how to rig the sails for a ship of the line from 1806, its a historical curiosity, but not how we do it now. I personally don't believe in the strong singularity, rather the arrival of powerful technology in lumps, filled with problems and always
    via bigdickzillionaire
  • depending on the intervention of skilled humans... but if you do thing that RSI with strong singularity happens, then for that scenario, the value of of the Al knowledge in the brains of men goes to zero very quickly.
    via bigdickzillionaire
  • Mus_Rattus The amount of time between "I know how to use Al and you don't, therefore I have an insurmountable advantage" and "Al no longer needs me to prompt it, therefore we're all in the same boat" may be shorter than you think...
    via bigdickzillionaire
  • Cheezburger Image 10671636480
    via Igor Omilaev

    We have to learn as much as we can about AI

    Licensed media provided as editorial context

Want More? Follow Us and Add Us as a Preferred Source on Google.

Tags

technology workplace artificial intelligence
Scroll Down For The Next Article