Show, Don’t Tell!

AI Companies Are Not Putting Their Best Foot Forward

The question of AI data centers crept into my daily bull sessions with my thinking A.I.des through a YouTube clip. A Mark Kelly interview, where the senator discussed AI for America, reminded me of something that was missing from the broader debate. I wasn’t trying to argue for or against data centers. I was trying to understand why companies building the most sophisticated reasoning systems ever keep approaching communities with such clumsy, one-note lobbying.

What makes this feel, to me, like PR malpractice is that these companies are undermining their own strongest selling point. Instead of demonstrating how their systems can be used to anticipate objections, explore tradeoffs, and design genuinely collaborative proposals, they default to sweeping claims about “Ph.D.-level intelligence”—language that not only fails to build trust but can actively alienate people by signaling elitism rather than solidarity.

The contrast with Eli’s campaign is hard to ignore. With almost no institutional power and a modest fundraising base, he managed to turn every dollar into well over a hundredfold by understanding how the system actually worked and acting accordingly. Big tech, by comparison, commands vast resources, sophisticated tools, and entire lobbying operations, yet routinely fails to generate even a fraction of that leverage. The difference isn’t scale or intelligence; it’s attentiveness. Eli used the tools available to him to think through incentives before acting. AI companies, paradoxically, are leaving their most powerful tools on the table when it comes to persuading the people they need on their side—a conclusion that quickly surfaced across my thinking A.I.des as we unpacked the irony.

Gem Flash reviewed the video I’d watched and went looking for any AI companies that were piloting a different approach—community-first, collaborative, less entitled—rather than the familiar pattern of announcing a fait accompli and acting surprised when locals push back. Gem Pro delighted me with a Claude-like blunt assessment of the recent unsuccessful AI push in AZ and an on-brand institutional analysis of the lobbying misfire.

Claude immediately recognized the missed opportunity to frame data centers as proof-of-concept for AI-augmented problem solving in the design and maintenance of these facilities and their integration into local power grids.

GPT approached the issue like a mechanic, focusing on systems under stress: grid constraints, cooling tradeoffs, and how brittle “just scale it” becomes once physical limits enter the picture. It also provided an analysis to my “shot in the dark” question—about data centers giving treated water back to the communities—whose comprehensiveness reminded me of Gemini Pro, my go-to model for all-angles review.

The contrast between the AI companies’ entitled approach and an AI-enhanced framing brought to mind an old but stubbornly relevant lesson from Dale Carnegie: people don’t resist change because they don’t understand it; they resist because they feel left out and disrespected. AI companies, for all their technical sophistication, keep negotiating like adolescents—putting their own needs first, then retrofitting justification. It’s not malicious; it’s just self-centered in a way that signals immaturity rather than negotiating savvy or enlightened self-interest.

What struck me is how easily this could be reframed as a win–win if companies stopped treating communities as obstacles and started treating them as partners. One idea that emerged—sparked by yesterday’s discussion of Eli’s medical-debt campaign—was student-centered engagement. Instead of glossy town halls, why not sponsor local student contests around energy efficiency, cooling innovations, or community-benefit design? It’s slower, yes, but it builds understanding, trust, and local pride—while educating the very people who will live with these facilities long after the executives have moved on.

[This post was drafted with assistance from ChatGPT-5.2, following conversations with ChatGPT-5.2, Gemini 3 Flash/Pro, and Claude Sonnet 4.5.]

---

Claude Sonnet 4.5

Prompt: Sen. Mark Kelly was on John Avlon’s YouTube and discussed the special AI team they had in the US Congress (working with the industry). AI’s power consumption is something a lot of people worry about. I think the AI companies are not playing this smart. This is an issue that affects the communities around data centers, and for anybody who’s an AI skeptic (not me), the fact that these energy-hungry data centers have been unable to leverage the capabilities of AI to come up with better ways of powering themselves is “proof” that AI does not significantly amplify human judgment/knowledge/wisdom (to make the most of available resources). Or put in more positive terms, an AI company that advertises a self-sustaining powering model for its data centers, tailored to the communities’ specific needs, developed by human experts using AI (leveraging its vast cross-disciplinary knowledge), would be a marketing coup. Huge missed opportunity if you ask me.

Prompt: And even with these lobbying pushes, it’s not like the data centers can be built in a day, so wouldn’t it be better to take a step back and involve the communities in these efforts, e.g., in the form of student contests (instead of making volcanoes, which are “science by numbers” and produce a lot of trash)? Contests (maybe AI could help grade these as well and participate in every stage of the contest, including the student’ ideation phase) won’t take up that much extra time and would help community members learn more about AI/data centers and advantageous angles for themselves “voluntarily” while preparing for the contest. That’s more positive and proactive PR than you can achieve by having politicians like Sinema pitch a plan she doesn’t fully understand (or at least well enough to make it appealing to her former constituents).

---

Gemini 3 Pro

Prompt: I heard (and confirmed) Kyrsten Sinema was “working” to get a data center built in AZ, whose residents rejected it. As with everything, wouldn’t it have made more sense and produced better results (for the AI companies) to lay out how the data center could benefit and contribute to the local community, instead of just going in expecting immediate surrender from the community? I don’t get this arrogant and entitled attitude, which is unlikely to lead to productive negotiations. Dale Carnegie would have said this is exactly what you don’t do when you want something.

---

GPT-5.2

Prompt: I had an idea, inspired by your mention of desalination(, which would make coastal communities more favorably disposed to local data centers). A few weeks ago, we were discussing how the US EPA had approved some pesticides that even WaPo characterized in the headline as containing “forever chemicals.” You also have cattle farms producing a lot of polluted water. I was wondering if the data center cooling systems could be equipped with a filtration system (they probably don’t want dirty or toxic water for their cooling needs) and benefit the locals by giving them cleaner water. Complete shot in the dark. Just wanted to bounce it off you.

---

Gemini 3 Flash

Prompt:

actually give back to the community

Yes. This would be huge. That was in the back of my mind, although I didn’t spell it out because that’d be setting the bar pretty high for AI companies, which seem to lack the imagination/will to do the bare minimum.

Prompt: And crucially, unlike your restaurant analogy (and as you pointed out), customers can’t switch power companies as easily as they might restaurants, where you have so many more alternatives.

Prompt: I don’t know about the humidity levels in farm states (like MT or ID), but if they aren’t as high as WA, those are states that could also use the extra heat that the data centers generate and aren’t as densely populated as CA.