Here at Hightower HQ, we’ve been spending a lot of time reading about and supporting the rebellion against AI imperialism and the data centers that come with it. We’re also seeing a lot of misinformation and Chicken-Little-style scare tactics being deployed by people for and against AI, and since I’m a nerd who doesn’t like to toss the tech baby out with the tech bathwater, I thought we’d take a moment here to offer a brief explainer of what AI is and is not.
In part two, coming next week, we’ll talk about data centers: what exactly they are, their impact on the power grid, water supply, environment and economies of the communities they’ve moved into.
✊ Fund independent tech truth →
AI stands for “artificial intelligence,” which is a broad term that’s often misused. Overall, “artificial intelligence” encompasses a number of technologies that are designed to handle massive computational problems thrown at them.
The current big hubbub, though, is about a more narrow type of AI called “generative AI.” This is a type of technology that can be used to generate stuff: humans can use these kinds of software as tools to make text, images, computer code, and much more.
But we can get even more specific! Most of the current conversation centers on “large language models,” or LLMs, which are a subset of generative AI. Think ChatGPT (made by OpenAI), Claude (made by Anthropic), Gemini (from Google) and more. These technologies have been built on large amounts of language (books, internet websites, and much more, often without the consent of the original authors and artists—including both Hightower and me) to train them to behave like and model (mimic) human language.
A tiny, common example: have you ever been texting on your phone, and on the screen, the phone tries to help you by predicting the next word you’re going to type? That’s basically what LLMs like ChatGPT and Claude are: extremely complex and sophisticated predictive text. They are not conscious and “thinking;” they are analyzing what you’ve given them, looking for relevant patterns and knowledge within their learning and online, and then spitting out responses that are (hopefully) relevant to the prompt you’ve entered.
Other types of AI, besides “generative,” include:
Predictive (looking at a set of data and forecasting future outcomes). Example: Your credit card company alerting you to potential fraud when you buy something outside of your usual patterns.
Diagnostic (detecting patterns or anomalies in, for example, medical environments). Example: Radiology AI scanning a mammogram and flagging a small mass for review by a human. This actually happened to me: I had 2 very small masses that were noticed by AI, but not humans. The radiologist confirmed they were worth biopsying, and while they both turned out to be benign, I was grateful to have them flagged.
Agentic (planning and fully executing multi-step tasks). Example: You tell an AI assistant to “book me the cheapest flight to Austin next Thursday and add the info to my calendar,” and it does so, without you doing each step yourself.
Each of us here on the Hightower team has a different take on using AI. Hightower never uses anything outside his own brain to generate his writing, of course—everything you get is 100% Jim Hightower, no bots involved. Laura avoids using AI whenever possible. I use Claude to brainstorm ideas in my creative work, to help me write computer code for my projects, and to diagnose & fix tech problems.
Partly, it’s because we now have hardware that can handle the massive computational power that these technologies require; a lot of this stuff has actually been around and in use for a long time. Spam filters on your email? AI. Siri on your iPhone? AI. Google Maps rerouting you based on traffic? AI.
Generative AI like ChatGPT and the kinds of party tricks it can do attracted a lot of attention in the last few years, which really brought the idea of AI into the mainstream.
But! The “explosion” of AI (in the sense that every piece of software we use in our lives now seems to have an “AI assistant” of some kind) is not because the technologies are now so good that they can be applied to everything, or are otherwise an inevitable part of all software. This is where things become political.
Ever since the general public has become more familiar and somewhat comfortable with the idea of AI, many, many companies are feeling pressure from their leaders and investors to bet on its usefulness. Typical capitalism trying to “keep up with the Joneses”: “If Acme has invested eight billion dollars in this tech, we better, too!” (As someone who has literally never understood why our economy is entirely run on vibes, aka, the stock market and coked-up traders making decisions that affect us all depending on their whims and moods, this makes me insane.)
Within capitalism, corporations are not allowed to plateau or coast: they must constantly show that they are growing, their profits are increasing, and their value to shareholders is increasing. Investing in AI is a shiny way to do that, even though this investment is still entirely unproven as a profitable move. (Not to mention the unresolved ethical and safety issues still at hand with generative AI.) The companies building and running the leading AI models (ChatGPT, Claude, etc.) have yet to show that there’s a way to make money that matches or exceeds the enormous level of financial investment they’ve made. The economics of actually running these models are brutal, and the gap between revenue and computational costs hasn’t closed. And yet, bajillions of dollars are being poured into the hope that it will, somehow, all work out. Sound like a bubble to you? Yep, us too.
What worries me most in this particular corner of the AI debates is that there are only a handful of companies running the table. Google, OpenAI, Meta, Amazon, Microsoft and Anthropic already have a stranglehold on the sector, and are using this power politically to prevent technological and economic guardrails from becoming law. They argue the same old, same old “But China! But Russia!” as a way of scaring legislators to leaving companies unregulated.
In part 2, we’ll talk about data centers: what exactly they are, their impact on the power grid, water supply, environment and economies of the communities they’ve moved into.
Here are some articles and podcasts I’ve read and listened to recently that have had nuanced, useful conversations about AI and the implications on workers, the environment, artists and more:
Life With Machines by Baratunde Thurston. Baratunde is a longtime homie of mine, who shares my deep love and enthusiasm for technology, but who also distrusts the exploitation of power that can come with it.
Dr. Chris Jones wrote a several-part series on AI for his newsletter: start with the first post here. He’s also running for Congress in Arkansas and is kicking ass— support his campaign!
How Does AI Mess With Our Creativity? An episode from the Culture Study podcast featuring author Vauhini Vara.
An author’s $2 million book deal was pulled over suspected AI use. Where does publishing go from here? A conversation with three writers on AI usage that contains a lot of nuance that’s often missing from AI and creative conversations.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.