We Need A COP-bot

Why Climate Action Needs a Feedback Engine

I’ve written here about next-generation governance, suggesting approaches based in both AI trends, and in the 20th-century science of cybernetics (which is not the same thing as information processing). Theory’s all very well, though—what about concrete examples?

I’m trying to bend my curve away from the theoretical, towards the practical, so in this post I’ll give an example of alternative governance that’s both possible, and desperately needed.

As I’m writing this, COP30 is going on in Brazil. It’s hard to see what’s going on from the outside; today’s headlines were all about protestors breaking in and injuring a couple of people; while that’s a newsworthy event, it’s not the point of what’s going on down there. What I want to know is, what progress has been made today, who’s leading and who (other than the US) is trying to drag the process down? More than that—like probably billions of people, I want to feel that there’s some way I can contribute, not just to the execution of the policies that result from the conference, but in the shaping of them. That’s what those protestors are trying to do. Well, they’re there, on the ground. But what about the rest of us?

Two Steps Forward?

The Paris Agreement was meant to create an the kind of cybernetic feedback loop for nations to govern the energy transition. It replaced top-down targets with a cycle of self-defined Nationally Determined Contributions and periodic global stock-taking. This is not working very well, partly because the loop stops at the national border. There may be mechanisms internal to specific nations to keep them on track, but no corresponding control system exists at the global level.

In the language of cybernetics, the system Paris created is (at the transnational level) a top-down control apparatus; the loop of feedback from the bottom back to the top is not closed. What’s more, at that transnational level, the control system has no authority to enforce existing or new targets.

Despite near-universal participation, the Paris goals are slipping out of reach. The first Global Stocktake concluded that even if every national pledge were fulfilled, the world would still overshoot 1.5°C of warming. The failure is not just political but structural. Governments set long-term promises that depend on short-term decisions no one wants to make. Fossil infrastructure keeps being approved even while net-zero targets are being heralded; you can see that in Canada’s new budget, which invests heavily in natural gas and maintains government subsidies to the fossil fuel industry that frankly should have been removed decades ago.

Investment in green technologies remains an order of magnitude below what the transition demands; ironic, considering that even without it, renewables are eating oil’s lunch. Where ambition and funds do exist, deployment issues remain; China is still building coal plants, even though they don’t intend to use them; ostensibly they’re there as a ‘backup’—’just in case.’ In case of what? The vastly distributed renewables grid is far less vulnerable than these isolated, easily targeted behemoths.

Beneath these visible problems lies a subtler one: you and I can’t see, day to day, how our own choices connect to abstract international goals. The climate story remains framed in parts per million rather than in daily life.

Enter the COP-bot

If you can find a copy of Stafford Beer’s book Diagnosing the System For Organizations, grab it—it’s rare but priceless. In this short manual, written for managers, this classically eccentric British cyberneticist gives his most compact and easily digestible definition of viable systems. One of his points is that such systems are always embedded in, and contain other viable systems; structurally, they’re fractal.

What I’m calling the COP-bot is a “collective intelligence” viable system. I’m using the terminology of viable systems to indicate that it’s not one thing, but many that that operate on multiple levels and that each subsystem is independent and able to survive on its own; and I’m giving it the collective intelligence label to point out that this is not an “AI overlord” or centralized application, but a set of interlocking institutions and practices. Their purpose is to ensure that the world stays on track to eliminating CO2 pollution. There are community-level systems that you can participate in, and these are embedded in national ones that are in turn components of a global feedback loop.

The book to read to understand Collective Intelligence

Within each viable system, we leverage AI in such a way that it does not negotiate or decide but observes, facilitates, and clarifies. Its purpose is to make the climate governance process legible, coherent, and responsive without altering its political balance. At the COP level, it could track negotiation texts in real time, every new bracket or deletion mapped against prior drafts. Historical precedents from past COP decisions could be retrieved instantly, letting delegates see where previous compromises lay instead of rediscovering them through re-negotiation. The thicket of acronyms and cross-references that clogs every COP could become transparent, with each paragraph linked to its legal lineage and scientific context.

An institutional COP-bot could also translate adopted decisions into structured, machine-readable “goal objects,” making every commitment traceable by indicator and timeline. It would serve as a memory, ensuring continuity across years and presidencies, and as a magnifier, giving smaller delegations the same informational reach as the large ones. For journalists, observers, and researchers, it would open the process to genuine scrutiny rather than rumor.

If such a system wasn’t embedded in another, higher level one, and if it did not contain lower-level equivalents, even this kind of reform would remain one-directional. It would make the COP more efficient and transparent but still fundamentally top-down. To complete the loop that makes this a piece of truly democratic cybernetics , we have to connect the deliberate, national level directly to households, cities, and communities. The national-level system also needs to be able to stream its live data up to the global-level process.

Personal and Civic Agency

If the middle-layer COP-bot listens to governments, the ones embedded in it must be governed by lower-level stakeholders, including people like you and me. These bots interact with organizations and people in homes, classrooms, and workplaces. It’s at this layer that the idea of using AI for collective intelligence is most clear. Here, artificial Intelligence is used to translate abstract international goals into concrete local policies and actions. Where the institutional system tracks negotiations, this one would track what citizens can do, what stands in their way, and how those efforts collectively shape national performance.

This public or personal COP-bot would be less a policy instrument than a tutor, a navigator, and a mirror. It could take the high-level outcomes of each COP—the emission targets, finance mechanisms, and technology pledges—and render them comprehensible at the scale of a neighborhood or a home. It would turn commitments into choices, showing not what must be done in theory but what can be done here and now. A heat pump installation, a local transit upgrade, a change in dietary habits—all of these would become visible contributions to the same global narrative that unfolds each November at the climate summit.

Such a system could also expose the friction points that don’t surface in the data. It wasn’t data that made rioters break into the conference site this week. And even at a bureaucratic level, say when you try to apply for an efficiency rebate and give up halfway through, that act of frustration is itself valuable information. When a municipality lacks the authority to implement a policy that its national government has promised, the impasse should not vanish into silence. By allowing people and local governments to record their efforts and obstacles, the public COP-bot would implement one of Beer’s most cunning insights into viable systems: that the top and the bottom have to have a communications channel. Once again, the system at this level would not prescribe behavior. It would be there to reveal or make connections, so that the web of interdependence that links individual agency to planetary outcomes is visible and tangible to stakeholders at all levels.

The lower-level COP-bot would be both an educator and an amplifier. It would teach climate literacy through action, turning technical language into civic understanding. Together, the two systems—the institutional and the personal—form the beginning of a genuine democratic feedback loop for climate action.

Closing the Loop

Although the UN does not possess a world government’s coercive authority, the Paris architecture is supported by institutions and agreements that can shape national behavior. At this level, the COP-bot would function to make these institutions more capable of fulfilling the enforcement roles they already possess.

The most immediate beneficiaries are the bodies housed within the United Nations Framework Convention on Climate Change. The Secretariat and its subsidiary organizations—particularly the transparency and implementation bodies established after Paris—are charged with reviewing national progress, verifying data, and identifying gaps. They operate today with limited staff and slow reporting cycles. The institutional COP-bot would give them continuous, machine-verified streams of nationally and sub-nationally sourced climate data. Functionally, this is very similar to the Cybersyn “dashboard” Beer and Allende implemented in Chile in 1970. Instead of relying on occasional reports that may be incomplete or politically filtered, the Secretariat would have a digital twin, on the planetary scale, that shows how nations are actually performing with regard to their commitments.

While the UNFCCC cannot punish noncompliance, it exerts real pressure through reputation, diplomacy, and agenda-setting. When shortcomings are obvious to every other state, the political cost of underperformance rises sharply.

Alongside the UNFCCC, the Article 6 mechanisms—the international carbon market rules agreed under Paris—become significantly more enforceable. These mechanisms depend on accurate accounting and transparent tracking of emissions reductions across borders. A COP-bot infrastructure capable of tracing the provenance of emission reductions down to local or project-level data would make fraudulent or exaggerated carbon credits much harder to pass off. In this way, the market itself becomes an enforcement tool: credits from nations that fail to maintain integrity simply lose value, and participation in Article 6 trading becomes conditional on maintaining verified datasets. No police force is needed; the market punishes noncompliance by refusing to accept tainted credits.

Beyond the climate institutions themselves, the multilateral development banks — the World Bank, the regional banks, and the Green Climate Fund—already influence national climate policy through the terms of financial assistance. They do not enforce climate obligations through threats; they do it through conditions attached to concessional loans, grants, and investment guarantees. If these institutions are supplied with precise, real-time data from the COP-bot systems, they can tie financial disbursement directly to verified implementation milestones rather than to promises on paper. In practice, this turns climate action into a prerequisite for accessing billions of dollars in affordable capital, creating a powerful incentive for follow-through.

The trade system adds a different but equally strong kind of enforcement. Nations that lag on climate mitigation increasingly face carbon border adjustments—tariffs or levies designed to compensate for carbon-intensive production. These measures rely on consistent emissions accounting and supply-chain transparency. A global COP-bot infrastructure that standardizes and verifies emissions data makes the application of such trade adjustments much more precise and much harder to challenge legally. As a result, trade access becomes indirectly conditional on climate compatibility. The enforcement agent here is not the UN but the global marketplace, operating according to rules that become sharper and more defensible once data integrity is no longer in question.

Financial regulation adds yet another layer. Central banks, securities regulators, and international standards bodies like the IMF, OECD, and IFRS Foundation increasingly treat climate performance as a factor in financial risk. They influence trillions of dollars of investment by defining what counts as a green asset, how firms must disclose climate exposure, and how credit risk is assessed. If these regulators can draw on reliable, automated, cross-jurisdictional data—the kind the two COP-bots would produce—they can adjust capital reserve rules, risk assessments, and disclosure requirements in ways that reward genuine decarbonization and penalize delay. In this sense, enforcement arises from the cost of capital itself: the price of borrowing increases for laggards and falls for leaders.

Scientific and anticipatory institutions like the IPCC, UNEP, WMO, and IEA benefit as well, because the intelligence they generate becomes more accurate and more rapidly updated when grounded in verified ongoing data rather than in slow national inventories. Their assessments increasingly influence everything from trade negotiations to investment strategies. Once their models are tied to real-time feedback loops, their influence on national policy becomes more direct, because governments rely on these assessments when setting their own plans and when interacting with international partners.

What ties all of these together is that none of them enforce through coercion. They enforce through consequence, through conditionality, through reputational pressure, and through the invisible leverage of finance and trade. The COP-bot layers transform climate action from an intermittent, self-reported process into a continuously monitored and publicly legible regulatory framework. By doing so, they strengthen the ability of existing institutions—not new ones—to coordinate global behavior.

In this way, the global feedback loop becomes concrete: the institutional COP-bot turns international decisions into structured commitments and tracks them rigorously; the public COP-bot reveals what is happening on the ground; and the transnational institutions that already shape the world—financial, diplomatic, scientific, and commercial—use this newly reliable information to reward compliance, penalize delay, and gradually steer nations toward the trajectories they have already agreed to follow.

Devilish Details

This is just a general sketch—one of my high-level design provocations, rather than a detailed description of how the system would function. If you’re interested, I can provide such a detailed design, though I admit I’ll have to lean heavily on AI myself to flesh out the design. (I’ve become used to doing that and understand the limitations and benefits.)

What do you think? Can we harness AI for a collective intelligence application—one whose components are primarily human beings and existing institutions—to break through roadblocks and unstick the stuck in the COP process? I’d love to get your take on the idea.

—K

---