NIO found something fascinating. It’s called a Civil Rights CAPTCHA. The idea is to filter spam-bots by posing an ideological question that functions as a test of humanity. The implications are truly immense.
The fecundity of Alan Turing’s Imitation Game thought-experiment has already been remarkable. It has an even more extraordinary future. The Civil Rights CAPTCHA (henceforth ‘CRC’) adds an innovative twist. Rather than defining the ‘human’ as a natural kind, about which subsequent political questions can arise, it is now tacitly identified with an ideological stance. Reciprocally, the inhuman is tacitly conceived as an engine of incorrect opinion.
Even the narrow technical issues are suggestive. Firstly, the role of the spam-bot as primary Turing test-subject is an unanticipated development meriting minute attention. It points to the marginality of formal AI programs, relative to spontaneously emergent techno-commercial processes (whose drivers are entirely contingent in respect to the goals of theoretical machine-intelligence research). Due to evolving spam-onslaught, many billions — perhaps already trillions? — of imitation games are played out every day.
Spam is a type of dynamically-adaptive infection, locked in an arms race with digital immune systems. Its goals are classically memetic. It ‘seeks’ only to spread (while replicating effective strategies in consequence). Clearly, the bulwarks of visual pattern-recognition competence are already crumbling. As a technical solution to the spam problem, CRC makes the bet that tactical retreat into the redoubt of higher-level (attitudinal-emotional) psychology offers superior defensive prospects. Robots are expected to find humane opinion hard.
By taking this step, CRC establishes a new class of agents — based on moral incompetence. The demonstration CAPTCHA text has been carefully selected to elide the element of ideological decision (while simultaneously, and strangely, foregrounding it): “In 2011 the freedom of the press was strengthened in Moldova, following a general improvement of the legal and political situation in the country,” it states, asking: “How does that make you feel?” The response options are “Tame”; “Crushed”; or “Hopeful”. “Tame” seems closer to grammatical error than crime-think, but between “Crushed” and “Hopeful” there is an obvious political choice. (It is this that NIO picks up on: rogue AIs and Putinists need not apply). The ambiguous invocation of ideo-emotional competence is compounded by the explanatory text:
A CAPTCHA is a test to tell wether a user is human or a computer. They mostly come in the form of distorted letters at the end of comments on news sites, blogs or in registration forms. Their main function is to prevent abuse from “bots” or automated programs written to generate spam. Civil Rights CAPTCHA is unique in its approach at separating humans from bots, namely by using human emotion. This enables a simpler and more effective way of keeping sites spam free as well as taking a stand for human rights.
A “stand for human rights” in this context is an argument that has finished with arguing, and seeks instead to install itself as a mechanical permission protocol. This is the “algorithmic governance” of the Left. As things get rougher, it will grow.
ADDED: Nydwracu deserves credit for the first catch (I’m confident he’s too magnanimous to care).