AI Amplifies You
As a father of a 12 year old, I'm looking for ways to help them get interested in AI and technology. I've written a script called 'drinkme' to set up a Linux server, and I'm creating a Pipulate server with a Jupyter Notebook for foundational tech skills. My goal is to help my child survive the coming AI craziness and learn real skills and knowledge. Come explore how AI can be used to help your child!
AI-Powered Learning: Helping My Child Prepare for the AI Revolution
By Michael Levin
Thursday, March 2, 2023
If you are a cheater, AI will help you cheat. You can cheat in a statistically optimized way and achieve your goals of perfect cheating. But if all life is really just a bunch of layered-up survival tricks over a background of hilarious shenanigans as I suspect, what’s the difference between AI-assisted cheating versus some other less loopholey path in life in which you actually have skills and know what you’re taking about?
Well if your mere goal in life is to get to the end of the game with some 60 to 120 years behind you, then there’s really no difference. Cheat away! You can live a life of non-deliberate chaos, hiding your insecurities behind an AI-induced air of false confidence, managing to get through it all, managing to dump the hard work on other people, and now AIs, and avoid personal growth indefinitely. It’s a workable life choice.
Now that there’s an AI-assisted cheater’s path through life, you can avoid gaining any marketable or soul-feeding skills, relegating yourself to the ranks of the AI-dependent. You will be dependent on AI products without the guilt of having to go to Mom and Dad all the time, but it’s playing the exact same role. It’s disabling you as a functional, versatile, and I would argue satisfied and fulfilled adult human being.
We spade and neuter our puppies and kittens keeping them in a playful but dependent state. They’re easily controlled and can themselves be turned into products to be bought and sold. Cats take up unnatural meowing sounds because it comes across like a baby crying and gets us to do for them. Just put us in the role of pets and AIs in the old of corporate owners to get a sense of what’s being put in place as you read.
This for example is most definitely not written by an AI large language model. It couldn’t be. Can you sense the difference in tone? Chances are, AI never trained on me because I’m not in the mainstream. I write mostly just for myself to think out loud, refine my thoughts and to guide my day-to-day choices and help me hone my skills. It’s not on Medium.com. It’s not ((crawlable) on GitHub.
And even if AI did train on me, I’m not in the mainstream. I’m a statistical outlier and my opinions will never reach the statistically relevant whitewashed AI responses. Stuff like mine you’ve still got to get from original source, and I don’t really care whether you do or not because I make the same amount of money whether you read this or not. It’s 100% altruistic and self-improvement, and thus of zero interest to the economically incentivized content machine.
Sure my writing could use a lot more revision and refinement to be more mainstream and AI-consumable. I should probably be trying to push it into every publishing channel for exposure, audience acquisition and influence. But it is a strength that I don’t have to. It’s a strength that my priority in writing is really just to give my brain a workout digesting and processing all the new input.
Garbage in, garbage out. Humanity in, humanity out. That’s why AI’s giving us. Not the best of us. Not the worst of us. But only the most effective of us in getting our messages out, crawlable and deemed authoritative by whatever algorithmic processes. Throw it all in the blender, smooth it up and dole it out to the world in an AI smoothiegasm that SciFi fans have been drinking for decades.
Most things published today is to trick you into clicking ads so they can make money. The creation of that content was not genuinely incentivized to make your life better. It was incentivized to make you somebody’s customer. It was incentivized to make you an obedient child. You know the primary skill of a child trying to get what they want without putting in the work? Wahhhh! That’s the sound of a baby. That’s the sound of get-rich-quick cheaters prompting AIs.
You can say wahhhh in a rudimentary spoken-language. This is like a long wining meow in cat language. As humans, we pick up our native spoken tongue naturally from our parents, culture and society. It doesn’t take much work to talk. But it does take work to read and write. This is why nearly all humans can talk, but not all can read and write. Literacy takes that extra bit of work and is the responsibility of parents to put children on the path to literacy.
I am the father of a 12 year old. Any sentence with the words math or programming in it triggers the kid like they were trained to hate both. But they’re learning to video-edit, so programming is being snuck in. I’ll look for ways to amplify THAT. And I’m hoping they’ll get swept up in the AI revolution and get curious about it and come ask ol’ Dad. I think it genuinely surprises them when they see my bona fide tech skills when they’re actually interesting or of use in their world.
So that’s the makings of a general plan. Get good at a variety of common everyday AI-skills. One must be able to get through a whole Fast.AI course. I like Jeremy Howard’s approach. It’s very accessible and very compatible with my Python way of doing things. He does however promote Google Colab or other Cloud Notebook platforms as a way of avoiding local tech issues. There’s validity to that. It’s an easy way to keep it all Linux without the user even realizing.
But we’re in a different day today. Any Windows machine can easily run Linux and I’m coming to realize pretty much any Mac too. And you don’t even need to specifically install a general virtualization product. If you’re going with Ubuntu, you can install it with their MultiPass installer. Honestly I think it’s a blend of VirtualBox and QEMU FOSS PC virtualization code cherry picked for an easy-install version of Ubuntu on Windows, by the people who brought you LXD and the Linux snap store. Canonical is making it very easy indeed to run Linux on Windows or Mac host platforms.
And so, Linux is now the generic plumbing of tech, pretty much any way you slice it. There’s going to be a lot of virtualization, containerization and other ization software you’ll run into along the way. Some day you’ll run Linux on the hard metal as we say here and there. If you run a Raspberry Pi for example, an ARM version of Linux is running directly on the processor without necessarily any hypervisors. But they are multi-core now, so you never know.
One thing you should know is that at some point, it’s a generic unit of computing knowing as an instance of relatively modern Linux. That is to say, the directory structure is pretty much the same for everyone. There’s few surprises anymore. It’s a long-in-the-tooth, well vetted FOSS (free and open source) boot kernel, set of system services and set of commands. It’s gotten even much better now with the systemd approach to managing system services
Wanna jump on the AI bandwagon? Start using somebody’s proprietary service and likely paying some $10+/mo for a subscription. If it’s not that it some similar spin on that which also requires paying somebody at least $10/mo. If it doesn’t meet that criteria, it’s not profitable to anyone to be doing. That is unless of course there were more altruistic motivations fairy’s and unicorns. They want your money. It doesn’t have to be that way. Step #1: set up some FOSS instance of Linux. But that’s way too much.
Enter your Windows laptop, plus running a magic little script that will possess it and make an instance of Linux server run in the background. You can start to do stuff with this. You can start to get to know truly generic, free, mainstream, integrated into everybody’s everything and thus timeless platform on which to do stuff. Invest your time here where the carpet can never be pulled out from underneath you. Even the AIs run on Linux systems.
What’s more, there’s a clever way of putting off ever having to do anything over on the Linux side, and still start to get the benefits of transitioning your tech skills over to Linux today. The trick is to run JupyterLab so that you can run Python in a browser tab Windows-side, Mac-side or whatever and in actuality, your Python code is being executed by an instance of Linux running under WSL. Better still, it’s running as a system service so you can start to poke around and see how all that works.
Things are completely different now that Microsoft ships ready to install Linux with the wsl –install command. Better still, there are scripts that can configure the Linux-side to be some sort of functioning server. My “drinkme” script is one such server setup program. Currently, I have it written for Windows given the sudden broader reach of Linux now, courtesy of Microsoft. Come that with all the ChatGPT and Bing-provided Python code. Well, if you need somewhere to copy/paste that Python code into, have I got a place for you.
Start out in a Jupyter Notebook. It’s something you just get to in your browser after casting the magic drinkme spell. I call it a Pipulate server. I own the domain pipulate.com and you can pip install pipulate, though it’s not fleshed out all that much yet. You and I together here are going to start down this path together. I haven’t built pipulate up all that much yet, so it’s a perfect time to jump onboard if you’ve been looking for something to do to survive the coming AI craziness. These are good solid foundational tech skills that resist disruption.