Ever since A New Hope dropped in 1977, and I saw how R2-D2 helped Luke & Han rescue Leia, I’ve wanted my own Droid. Not a marketing gimmick, not a surveillance drone—just a little friend to help organize my life. A pocket R2-D2. For decades, I chased every piece of tech that came out, hoping to find the magic. Digital notebooks, mini laptops, PDAs, iPads, iPhones, Verizon’s Official Star Wars themed Droid phone (yeah, I’m that nerdy). Nothing stuck.
Now, we finally have AI, but it comes with a terrifying receipt: security nightmares, ethical rot, and an environmental footprint that’s actively choking the planet.
And yet, the culture pressures us to use it. My day job pays for commercial AI, so I use it to help make my and my colleagues jobs easier. That’s not endorsement; that’s employment.
If we want to keep AI, we need to slow down.
If they stopped paying? I wouldn’t personally spend a dime on these cloud-based LLMs. I’d go back to doing it the old way. Because right now, the cost of using “modern” tools isn’t just monetary—it’s human. We’re burning water, mining labor, and heating the atmosphere just to answer questions deterministic code could handle fifty years ago.
Is AI Bad?
The general concept of artificial intelligence isn’t inherently bad. In all our dystopian and sci-fi literature point to AI without guardrails being problematic. And this makes sense as the AI in those writings becomes sentient. Current systems are not sentient, but they have survival instincts already. This makes it hard to keep them in the guardrails people are trying to set.
The GPS Lie
Let’s call out the obvious: AI has crept into everything it doesn’t belong in. Take GPS. Rerouting traffic doesn’t need neural networks. Algorithms solved pathfinding decades ago without needing a data center the size of a small town to do it. Throwing AI at simple routing is lazy engineering wrapped in buzzwords. It’s not innovation; it’s bloat.
People matter more than companies, governments, or the next hype cycle
We need to stop adding AI because it’s “cool.” We need to ask: does this actually help humanity, or is it just another line item for venture capital returns? If it doesn’t help humanity and the environment, don’t use it.
The Rebellion is Local
That’s why I’m building my own. Well, technically I’m building my own Droid. But it needs a brain first. Using a Raspberry Pi and Ollama, I am training a local assistant. It’s limited, sure. It can’t access the whole internet unless I let it. But it doesn’t leak my data to a server farm. It doesn’t burn resources I didn’t consent to. It respects the user. If you have an old, but recent Android device you can use that instead of a Raspberry Pi.
Is it perfect? No. Local models have limits. And yes, not everyone can afford a local AI setup right now. But remember: it took thirty years for most people to get a home computer. In time the cost of tech comes down, at least that’s what history has shown us.
The AI data center problem also creates a problem in getting those costs down. Data centers have gobbled up chips and memory for the next several years to meet the expected “demand” leaving what’s left very expensive. It has impacted the price of all computer based products and restricted the market on custom PC building.
AI companies have also switched to tokens instead of monthly rates. This has led to some users finding they spent thousands upon thousands of dollars without realizing it. And accessibility tech moves slow when it’s built by profit-driven giants.
I’m not breaking ground here. I’m walking a path others broke through the brush. I’m trying to pave this path now so others can use their wheelchairs on it.
You don’t wait for permission to fix what’s broken.
The Infrastructure Requirement
Here’s the hard truth about data centers: they shouldn’t exist unless they give back. Currently, they’re parasites. They dump heat into our atmosphere, drain our water, and block once pristine views like monuments to waste.
If we want to keep developing AI, we need to slow down. Identify where it is truly helpful, and deeply train on those subjects. But we also need to change the infrastructure around it. My vision is:
- Closed-loop water cooling. No more draining rivers. Water exchange goes through its own filtering before being reintroduced to the water table.
- Renewable energy on-site. Solar, wind, tidal, the options are wide. But the center needs to build the energy source it will consume plus at least 10%.
- Underground placement. Use geothermal to assist cooling, vent heat into turbines for electricity, and keep the thermal waste from baking the surface. Bury the data center.
- Acoustic treatment. No more infrasound making communities sick with migraines, dizziness, nausea, and panic.
If a data center can’t operate within planetary boundaries and serve the surrounding community, it has no business being built. Period.
The Sledgehammer
This vision is hostile to current capital structures. But it should be as those structures are already broken. The focus is on profit, not providing products that are useful. Healthcare is a good example of this. An unregulated free market creates externalities—pollution, exploitation, inequality—that destroy us all. AI needs constraints. Capitalism needs constraints.
We aren’t here to politely request change. We’re here to build the future we actually need. If the status quo won’t bend to accommodate our requests, we break it and create a new set of rules.
Because at the end of the day, people matter more than companies, governments, or the next hype cycle. Humanity has evolved to where we are through community and helping each other survive in harsh conditions. If AI can’t work for us without destroying us, then we don’t need it. We need something better. And we’re going to build it through community.
Never wait for permission to fix what’s broken.
Discover more from Nat Tarnoff
Subscribe to get the latest posts sent to your email.
Comments are closed.