tl;dr: Green AI can only matter if it moves beyond hype—toward truth, transparency, and regeneration within planetary limits.
Over the past months, hardly a day has passed without headlines about artificial intelligence: new models, astonishing breakthroughs, and dire warnings. At conferences, in boardrooms, and in policy debates, AI has become a kind of secular promise—of productivity, control, even salvation. And increasingly, it comes wrapped in a green label. That convergence between sustainability and digital technology is fascinating, but also deeply ambivalent.
We live in a time where artificial intelligence is being hailed as the next great savior – for business, for society, even for the planet. The promises are grand: Green AI will solve the climate crisis, make our economies efficient, and lead us toward a sustainable future. But let’s pause for a moment and ask: how much of this is actually true? And how much of it is, to put it bluntly, bullshit?
That may sound harsh, but it’s an important distinction. As philosopher Harry G. Frankfurt famously argued in his essay On Bullshit, a lie still respects the truth – because the liar knows it and chooses to conceal it. Bullshit, however, is indifferent to truth. It doesn’t care whether something is right or wrong; it only cares about effect. In an age driven by hype, attention, and the constant need to impress, bullshit becomes the default mode of communication.
Nowhere is this clearer than in the tech world. Silicon Valley’s prophets of acceleration – from Elon Musk to Peter Thiel, from Sam Altman to Mark Zuckerberg – promise salvation through speed, scale, and disruption. But when every problem is a nail, and technology the hammer, we risk flattening the world rather than fixing it.
The Ecology of Limits
Let’s start with the ecological foundation. The concept of planetary boundaries, developed by Johan Rockström and colleagues, defines nine environmental limits within which humanity can safely operate – from climate stability to biodiversity, from nitrogen cycles to freshwater use. Seven of these boundaries are already crossed. We are far beyond the safe operating space of our planet.
What does this mean for AI? Every model training, every data center, every chip fabrication consumes energy, water, and rare materials. When companies claim their systems are “CO₂-neutral,” we should ask: does that include the full life cycle? Are Scope 3 emissions – the hidden environmental costs of hardware, logistics, and disposal – even counted? Often, they aren’t. That’s not sustainability. That’s marketing. Or as Frankfurt might say: bullshit.
Even the argument of efficiency (“Each query is now cheaper!”) hides a rebound effect: the more efficient we get, the more we consume. It’s like replacing every bulb with LEDs and then leaving the lights on longer because electricity feels cheap. Multiply that by a few billion data queries, and the picture becomes clear.
The Society We Build
Sustainability is not just about ecology; it’s also about society. The Wedding Cake Model of the Stockholm Resilience Centre visualizes this elegantly: the economy rests on society, and society rests on the biosphere. Without stable ecosystems, societies crumble; without cohesive societies, economies collapse.
This matters for Green AI. Technologies that improve traffic flow, reduce pollution, enhance medical diagnostics, or strengthen education genuinely reinforce the foundations of that cake. They support social and ecological resilience. But when AI is deployed solely for profit – to push ads, amplify consumption, or extract attention – it erodes the very base it depends on. That, again, is the essence of bullshit: ignoring interdependence for short-term gain.
The Technology Trap
Since the 1950s, humanity has experienced what scientists call the Great Acceleration – an exponential rise in energy use, emissions, and material throughput. We are now in the Anthropocene, an age where human technology acts as a geological force. Technology is not neutral; it always carries an ideology. As Marshall McLuhan put it, the medium is the message. Or, to borrow a simpler metaphor: if you only have a hammer, everything looks like a nail.
Our tools shape not only what we can do, but how we see the world. When technology becomes the ultimate solution, we risk turning existence itself into a dataset – measurable, optimizable, and, ultimately, exploitable. That worldview is not inevitable; it is a choice. It reflects the values of those who build and fund these systems.
Green AI can take two paths: it can accelerate the same destructive logic of endless growth and efficiency, or it can help us shift toward a paradigm of sufficiency, regeneration, and transparency. The choice is ours.
Beyond the Buzzwords
If we truly want to move beyond the bullshit, we need three guiding principles:
1. Transparency – Every claim of Green AI must be verifiable: full life-cycle accounting, open data on emissions, independent audits. Without that, we’re in the realm of slogans, not substance.
2. Sufficiency – Just because we can build ever-larger models doesn’t mean we should. Small, targeted systems often achieve what’s necessary with a fraction of the footprint.
3. Regeneration – Efficiency is not enough. AI should actively contribute to restoring ecosystems and strengthening communities, not merely minimizing damage.
To bring this into practice, we can think of AI-enabled systems that support adaptive resource management, circular economy logistics, and long-term monitoring of biodiversity. Each of these initiatives requires collaboration across sectors—research, governance, and civil society—to ensure that technology serves as a feedback loop for sustainable learning rather than as an amplifier of extraction. When we talk about Green AI, it is not about producing new apps or tools; it is about redesigning relationships among humans, machines, and the natural world so that each technological step is accompanied by ethical reflection, institutional change, and democratic accountability. Only then can efficiency evolve into stewardship and regeneration into an ongoing commitment to the commons.
The Honest Future of Green AI
To build a sustainable digital future, we must demand truthfulness from our technologies. Green AI is not a miracle cure, but it can be a meaningful tool if guided by honesty, humility, and responsibility.
The philosopher Frankfurt taught us that the danger of bullshit lies in indifference to truth. Green AI beyond bullshit means restoring that bond to reality – acknowledging limits, embracing transparency, and aligning technology with the regenerative capacities of the Earth.
Because in the end, sustainability isn’t about bigger models or faster chips. It’s about systems change and moral responsibility – about a renewed relationship with truth, planet, and each other.
Let’s build AI for that.