AMD held its Advancing AI 2025 summit in San Jose, California, on Thursday, headlined by CEO Lisa Su’s lengthy keynote address. However, it was not what Su said that alarmed me.
OpenAI CEO Sam Altman joined Su on stage to wrap up the two-hour keynote, and their back-and-forth included one short exchange toward the end that now lives rent-free in my brain.
Referencing the recent ChatGPT outages, Su asked Altman, “Are there ever enough GPUs?”
Altman, whose OpenAI is an AMD customer, replied, “Theoretically, at some points, you can see that a significant fraction of the power on Earth should be spent running AI compute. And maybe we’re going to get there.”
Which sounds great for the AI industry. But what about the rest of us?
Just because we can, doesn’t mean we should
Okay, look. I’m not the biggest fan of AI usage. One of the few types of AI that I find myself interested in is Intel’s on-device custom RAG AI Assistant Builder. It’s a small language model that runs locally and is mostly useful for the kind of busywork no one has time for.
Cloud-based AI data centers fuel my nightmares. The current pace of Artificial Intelligence growth is unsustainable. While Altman claims that users shouldn’t be worried about ChatGPT’s energy cost, adding AI usage on top of other sources of environmental pollution puts more pressure on a planet that was in dire straits before OpenAI programmed its first chatbot.
AI is not just an industry built on destroying the foundations of its own existence, though that is certainly an economic crisis in the making. It’s also incredibly damaging to the environment and to the education of future generations.
AI-generated deepfakes are getting so good that most people can’t tell real footage from AI video, leading us into a “future of AI control.”
Without guardrails, unchecked artificial intelligence growth is very much an ouroboros.
AI has already negatively impacted most industries. Even if you accept that AI will eventually replace human workers in most roles, then who will buy the products? AI won’t be selling laptops, phones, or software to itself. Throw in the ecological and educational damage AI systems can cause, and it’s a bleak future we’re looking at.
Thankfully, large language model (LLM) AI systems like ChatGPT are becoming increasingly efficient over time. So they might hit critical mass soon, and we won’t need to reopen Three Mile Island to run virtual assistants. There are also plenty of companies looking into sustainability solutions for AI systems.
But is it worth it?
We still haven’t gotten the “better Siri” we were promised. Agentic AI just sounds like a new way to commit to total social isolation. There may never be a “killer app” for AI.
Are knock-off Studio Ghibli profile photos really worth the cost? Then again, we’ve still not learned from Microsoft’s notorious Tay.ai blunder from 2015.
So perhaps it’s time to make like a dolphin and leave.
