PROME’s Care Core
Values of a farmer
There was a survey out last week that showed AI is less popular than nuclear waste. Ya got that? Nuclear waste.
So now, when people ask me what I do, I say I build something worse for humanity than nuclear waste.
Actually, I don’t, and I’ll tell you why.
For a long time I thought “Silicon Valley” was the epitome of this thing called tech that I thought was the coolest thing humanity had ever created. I fell in love with the game. Of pure creation. But what I didn’t realize until more recently is that I always had a positive view of it precisely because I approached it from a place that was arguably pure.
This act of pure creation was inextricably linked to my naive view that everyone would use it for good. To help people. To help animals. To help the planet. At the end of the day, we all work for the planet, for the animals, for the people. Everything else is just made up human stuff that help us pass the time. Money, games, politics, tech, yada yada.
It came from where I grew up and how I was raised. In Iowa, surrounded by farmers. Looking back, I’ve lived, worked, operated, advised, and partnered with people and companies all over this beautiful planet. There are amazing, incredible people everywhere.
99.9% of the time, folks just like you and I get up in the morning, sip our coffee or tea, go to work or school, enjoy some downtime with friends and family, and don’t do much that harms others. But it’s that small 0.1% minority who do that dominate the news cycle and mess everything up for the rest of us.
For many years now I’ve been diving into the depths of my own core values, and how that influences the companies I build, invest my time, money, and energy into, and how I use to that to approach our amazing clients, partners, customers, and suppliers.
The older I get the more I realize I have the values of a farmer. We put ourselves aside to feed the world. We hang together during floods, during harvest season, and our primary battle is with nature and the weather. Working with mother nature, to grow things that benefit all the passengers on this spaceship called Earth. We’ve only got one.
It was never about money or power as the north star that drove us. It was a hard day’s work in the field, coming home stinky, sweaty, and hungry after a pure day’s work. Spending some quality down time reading math and physics books and discussing them with our family. And boy, do we love a good flick and bucket of popcorn.
But at no point did we hire a publicist to sing our virtues across this spaceship, or do rug pulls. “Just kidding, this corn isn’t real, thanks for the money suckers.” Yah, that’s just not us.
So, now, in the age of emerging tech, and the evergence of self-driving, artificial intelligence, robotics, spatial computing, blockchains, space internet, manufacturing, and mining, it affords us a chance to look within ourselves and get really clear on our collective core values.
Here’s the thought experiment.
Do you trust a free Facebook robot to babysit your kids and tuck them in at night when they’re currently going through a trillion-dollar lawsuit centered on childhood harms?
Remember, the core values of the person building the technology, and the company, gets embedded deeply into their product. And that product directly impacts your child’s and grandchild’s well being.
It’s clear from the survey that you don’t trust anyone in artificial intelligence because, again, it’s worse than nuclear waste.
So, instead of going on the podcast circuit, and being yet another talking head, we decided to do something about it. We started over a decade ago.
Yesterday, I spent my Sunday proving something that, as far as we know, no other big tech company, or AI lab has done. We did a search across the scientific research paper community (hi arXiv) and the safety blog posts of the leading companies.
They’re the smartest people in the world, after all, right? Right? Ehh…I’m not so sure.
Let’s talk about how these large language models work.
They start with an algorithm. The default starting point of that algorithm is a series of random numbers. Then they “train” this algorithm to change the random numbers into numbers that represent a massive amount of internet data. Blog posts, social media posts, company marketing materials, books, stories, films, etc. The algorithm can’t tell the difference between something real, something made up (like the story terminator), and the vile nature of the comment section or forums like reddit. Remember, it’s just a number that is linked to another number.
So what you get is a core AI model that deceives, harms, mistrusts, and contains generally bad actor behavior along with all the good stuff humanity has to offer.
Now that this behavior is in there, using language at its core, these AI model companies try to train this bad behavior out of it. They put in guardrails around the outside, and do other sorts of fancy mathematical and software backflips. This is where the word “harness” comes from. They’re trying to bolt on safety and sadly, core values, around the outside of what’s already there.
That’s like putting a serial killer in a tutu and thinking everything’s going to turn out just fine.
These AI companies all skipped right past the most important part. The root cause. No doctor looks at a broken arm and goes, “Ya know what, put a band-aid on that arm and it won’t be broken any more”. No, they gotta go in and add a metal rod so it heals correctly.
You don’t treat the symptom, you treat the underlying root cause.
So that’s what we did.
I figured out a way to start with core mathematical principles that represent strong core values like fairness, benefiting the self and the group without inadvertently harming others, and integrity. I figured out a way to balance them, working together, as a unified system at its core. And figured out a way to maintain this over time so it can’t be hacked into. This is the nature of strong core values. They can’t be broken.
In fact, they’re called strong core values because even in the face of danger, disaster, unpopularity, or bankruptcy, the values don’t break. They’re resilient to damage. But you only know what they are once you’ve been seriously tested.
Back to the algo…Once I had this concept working, I then trained one of these transformer architectures on it, before it learned language.
Math comes before language. That’s the innovation.
Because it’s been proven that however an AI model starts, it maintains these values over time.
So, these farmer core values that I hold so dear, are now part of a baby AI model.
One note: this does not represent the fundamentally different architecture of PROME’s Biologic Intelligence. I’m still using the legacy transformer architecture that all the LLMs use to prove this novel approach. PROME’s architecture is a conversation for a different day.
If you want to learn more about how we implemented strong core values into math, see the experiments that have been run, and read the research paper, you can do it on PROME’s website.
I’m not giving away the source code, for obvious reasons. Because I don’t trust the tech industry that completely annihilated people’s trust. The good people of the planet who are struggling to make ends meet and pay for groceries so they can survive without a credit card or Buy Now Pay Later layaway.
Stay the course y’all. We’re working on something to help solve this. But it’s gonna take some time. We only have so much bandwidth.
In the meantime, let’s compare this to the Silicon Valley elite, shall we?
There’s this company called, funny enough, “Safe Superintelligence”, which has raised $5 billion. That’s 5 billion with a B. They don’t even have a product. Their website is a single page with a couple paragraphs of text. They’ve been invested in by the most storied and admired tier one venture capital firms in Silicon Valley. And guess what? The CEO got poached by Mark Zuckerberg because they threw a bunch of money at him. Zuck almost bought the company but failed.
So, I ask you. Do you think this company is the smartest on the planet? Do they have the strongest core values on the planet? Do they have the core values of a farmer, trying to feed and protect the world? Or is their core value money and power?
We already know the answer. They chose money, not the people. Hard pill to swallow but it’s right there on TechCrunch for you to read all about.
So yah, it doesn’t surprise me that 8 billion people on this planet don’t trust Silicon Valley and compare its product to nuclear waste.
You know who I trust?
The Iowa farmers who’ve been feeding the world for a century without asking for recognition. You don’t see the farmers buying up land in New Zealand or Hawaii in case of the apocalypse like those billionaire “leaders” of Silicon Valley.
You know who I look up to?
I don’t want to be like Mike, or put tech leaders on a pedestal. I look up to the Iowa Farmers. That’s who I want to grow up to be like.
If everything fails, my BATNA is to bartend on the beach. I lived off ramen and tuna at the University of Iowa studying math because it brought me joy and fascinated me. I can do it again. I don’t care about the money, the silly power games, or the social media influence. I care about improving the “land” for the people.
So yah, any AI or robots I create, will most definitely have the strong core values of farmers, and it damn sure ain’t gonna have a north star of money or power. If it does, I’m out.
Values are a weighing machine. Eventually it catches up to us all.
Choose wisely.
—Sean
Strong core values.
