The Last Border Is the Mind

Tech-Yes-Logy “There is no neutral technology. There is only technology whose conscience has been declared, and technology whose conscience has been hidden.” I want to introduce, in this chapter, a term that has become central to the work of the Peace Artists Foundation and BIG SYSTEMS over the past several years, and which I believe deserves to enter the wider vocabulary of the contemporary conversation about technology. The term is Tech-Yes-Logy. The name is, deliberately, a play. It contains the word technology, but inserts at the centre of it the word yes. The construction is intentional. Tech-Yes-Logy is the proposal that we should say yes to technology — emphatically, without the reflexive fear that animates a certain strain of contemporary anti-technology sentiment — but that the yes must be a conditional yes. We say yes to the technology that has been declared in service of human dignity. We say yes to the technology whose conscience has been made explicit. We say no — or, more accurately, we say not yet — to the technology that has not yet declared its conscience, that hides its values inside its optimization functions, that operates as if it had no values at all. The term emerged in a series of conversations I held in the late 2020s with engineers, ethicists, and policy makers across several continents. The core observation that drove the conversation was that the standard contemporary framings — techno-optimism, which says yes to everything, and techno-pessimism, which says no to everything — were both wrong. Both framings treated technology as a single thing to be evaluated as a whole. The truth is that technology is not a single thing. It is millions of things, each with its own design choices, its own values, its own consequences. A blanket yes is irresponsible. A blanket no is also irresponsible — and a great deal more harmful to the people who most need the gains that responsible technology can deliver. The right posture is Tech-Yes-Logy: a posture of active discernment, in which we say yes to specific technologies on specific terms, with specific conditions, in service of specific outcomes that have been articulated in advance. The rest of this chapter is a sketch of what such a posture would look like in practice. The Five Conditions of Tech-Yes I propose, as the operational core of Tech-Yes-Logy, five conditions that a technology must meet in order to receive the yes that the framework is willing to grant. The conditions are not exhaustive. They are designed as a minimum threshold — a set of basic tests below which a technology should not be deployed at population scale, regardless of its commercial promise. The conditions are: Condition one: declared purpose. The technology must declare, in publicly accessible language, what it is for. Not in marketing terms. In purpose terms. What human problem is the technology attempting to solve? What human need is it attempting to meet? What conditions of dignity is it attempting to advance? A technology that cannot answer this question, or that answers it only in terms of internal commercial metrics (engagement, retention, conversion), has not met condition one. Condition two: declared values. The technology must declare the values that are embedded in its design choices. What does the system optimize for? What does the system trade off against what? What populations is it built to serve? What populations might be disadvantaged by its operation? These declarations must be made in public, in language a non-specialist can evaluate, and they must be open to challenge. The contemporary practice of treating these values as proprietary trade secrets is, by the standard of Tech-Yes-Logy, insufficient. The values can be commercially sensitive in their implementation; they cannot be secret in their direction. Condition three: distributional honesty. The technology must publish, alongside its rollout, an honest accounting of who benefits from its operation and who bears its costs. This includes not only the direct users but the workers whose labour is required to maintain it, the data subjects whose information has been used to train it, the communities whose economic or social structures may be reshaped by its operation, and the ecosystems whose resources may be consumed by its energy or material demands. A technology that produces large benefits for a small population at large costs to a much larger population has not met condition three, even if its commercial performance is excellent. Condition four: graceful retreat. The technology must be designed such that, if it is found to be doing more harm than good, it can be wound down without catastrophic damage to the institutions and populations that have come to depend on it. This is a serious design constraint. A technology whose deployment makes itself indispensable, such that its withdrawal would cause systemic collapse, has met an enormous threshold of responsibility — and very few contemporary deployments meet it well. The deliberate engineering of graceful retreat capacity is a feature, not a bug. It is the design equivalent of an aircraft that can glide if its engines fail. We should not deploy at scale any technology whose institutional analog cannot glide. Condition five: continuous moral audit. The technology must be subject to ongoing review by a body that includes representatives of the populations the technology affects, that has authority to require modifications, and that operates with sufficient transparency for civil society to evaluate its judgments. This is not the same as regulatory compliance. Regulatory compliance is a legal test. The continuous moral audit is a normative test, conducted by bodies whose legitimacy derives from the populations affected rather than from the bureaucratic procedures of national legislatures. These five conditions are demanding. They are not, however, unfair. They are, in essence, the conditions that any other safety-critical industry — pharmaceuticals, aviation, nuclear energy, food safety — already meets, at least in principle, in well-regulated jurisdictions. The contemporary technology sector has been operating, for two decades, with regulatory expectations far below the standard applied to other industries of comparable potential consequence. This is the anomaly that Tech-Yes-Logy seeks to correct. Application to Artificial Intelligence The application of these five conditions to the contemporary AI sector is, by my reading, immediate and concrete. A large language model deployed to assist with medical diagnostics, for example, would, under Tech-Yes-Logy, be required to declare its purpose (improvement of diagnostic accuracy in specific clinical contexts), its values (which classes of error it prefers to make, given that all diagnostic systems must trade off types of error against each other), its distributional honesty (who has access to it, who does not, what populations are likely to be over-served and which under-served), its graceful retreat capacity (what happens to the institutions that have come to depend on it if it must be withdrawn), and its continuous moral audit (who reviews its operation and on what timeline). A social media recommendation algorithm, similarly, would be required to declare its purpose (the connection of users to content that, on net, serves their flourishing — not their engagement, which is a different thing), its values (what the system treats as a positive signal and what it treats as a negative signal), its distributional honesty (whose attention is being optimized, whose is being mined), its graceful retreat capacity (what happens if it is determined that the system, on net, harms the populations it serves), and its continuous moral audit (who reviews its judgments and how). A facial recognition system deployed for law enforcement would be required to declare all five conditions, and the bar would, in this case, be set particularly high, because the costs of misuse are particularly severe. A generative AI system used in education would be required to declare all five conditions, with particular attention to the question of how its deployment affects the development of the cognitive and creative capacities of the students it is interacting with. In each case, the application of the framework is straightforward. The framework does not require novel mathematics. It does not require unaffordable infrastructure. It requires, primarily, disclosure and accountability — the discipline of operating in public, with declared purposes, declared values, and ongoing review. This is the standard already met by mature professions. It is the standard the technology sector must now meet as well. Why “Yes”? A reader may wonder why, in a chapter that imposes such demanding conditions, I have chosen to centre the yes of Tech-Yes-Logy. Would not a sterner term — Tech-Audit, Tech-Restraint, Tech-Caution — be more accurate? The reason for the yes is, in part, philosophical and, in part, strategic. Philosophically, I am a person who believes that the long arc of human civilization has been a story of expanding capability — capability to feed, to heal, to communicate, to imagine, to connect, to build, to remember. Technology has been, throughout that long arc, one of the principal mechanisms by which the capability has expanded. To say a blanket no to technology is to say a blanket no to one of the principal mechanisms of human civilizational advance. This is, in my view, an unwarranted refusal. There is too much good that responsible technology can do — for the children of the sun who lack access to quality education, for the elderly in the kingdom of ice who lack access to companionship, for the smallholder farmers who lack access to markets, for the rural patients who lack access to expert medical advice — to take the position of blanket refusal. The renaissance argument is, in part, an argument for technology, not against it. Strategically, the yes matters because the political coalition required to govern technology well must include the engineers, the entrepreneurs, the investors, and the technically literate publics who are, on the whole, enthusiastic about technology. A framework whose first word is no will not be adopted by these coalitions. A framework whose first word is yes — but a conditional yes, with declared conditions that can be argued with — can be adopted, refined, contested, and, eventually, institutionalized. The strategic yes is the precondition for the substantive conditions. This is, in essence, the same political wisdom that animates the better welfare states. The Nordic project is not anti-market. The Nordic project is yes to markets, under specific conditions, with specific accountabilities. The same posture, applied to technology, is Tech-Yes-Logy. I close this chapter with a sentence I would like to see, one day, inscribed at the entrance of every major technology institution in the world. “We say yes to the tool whose conscience has been declared, and not yet to the tool that hides its values inside its optimization functions.” That sentence is the operational core of Tech-Yes-Logy. It is the discipline our species has not yet imposed on the most powerful tools we have ever built. The imposition is overdue.