Turing; a "sufficient" threshold of intelligence: based on speed/immediacy: the imitation game. This condition has now shifted from a sufficient condition to a necessary condition of artificial intelligence. Nick Bostrom: super intelligence & super catastrophic risks - center for the future of humanity at Oxford. = theory of runaway superintelligence: application of information from one system (playing go) to another domain (perfecting a culinary dish), leading to AI that are good at building AI then incrementally improving through exponential growth & very soon humanity ends - cataclysm. What fail safes must be put into place to keep this intelligence within the Petri dish? The problem is a human one, to formulate a heutristic question in order to program an AI to think at the network level abstractions approximating the human.

Artificial Stupidity: throttling of the capacities of the AI to be better for the human user: computer chess = CPU is made stupid so we can engage with it.

Vs

Artificial Idiocy: not a dumb machine but refers to when the AI performs its programmed task to such an idiotic extreme that it's efficiency becomes idiotic: paper clip maximizer machine= the entire atomic substrate of planet earth is rendered paper clip. No bug/glitch, performed function perfectly to a catastrophic degree > ethics machines expanding to an idiotic degree.

Victor Tausk: "The Influencing Machines"

.....

On Automation in the Stack

Schi-morphism: one thing standing in for another, understanding the affordances of the previous thing: "throwing away" "documents" in the "trash can" in the GUI. The trash model does not reflect - is discontiguous - from how the info is read/stored/etc on the CPU hard drive. The problem with automated schimorphism on various layers of the stack: great & horrible that we don't have to deal with them all the time. AI at landscape scale: problem of automation: who what how is being automated & what it is being automated against: when does it solve a problem & when/how does it cause more problems or by increasingly complex without significant results - automated refrigerator.

Evolutionary robotics: "Intelligence without Representation" 1991 - models of AI inherited from anthropomorphic robotic design.

What is incredible about machine/AI utilizing Natural language is the implications for English itself - google translate AI making its own language to translate between languages.

Economic automation, or how automation shifts the relative risk patterns within the economy being automated. Things that used to be owned privately get pushed back into a service model : pushed to the periphery of the platform : the cascading effects of the democratization of mobility: takes the risk back from the center towards the edge. If we can see these benefits, what are the end points of this virtualization? City by city basis depending on the infrastructure. With automation we do away with the possibility of demand-side crash. Distortion of the pricing schemes of capitalist labor: "Inventing the Future".

Voice as an interface - specifically from human to AI. Not that it doesn't have a voice, but perhaps we are limiting it to merely a voice instead of all the possible ways it could express. The capacities may be less dependent on the capacities of the AI to articulate & more so on the limitations on the range of sounds that are available to articulate. Sounds that can be interpreted as intentions, a microphone personified. Conversational UI: not a composite users but appears like a discrete user.

Paro & sex dolls: emotional infill: more difficult to engineer an AI that humans feel emotionally connected to than it is for an AI to be connected to the human user. Through these interactions we come to train each other. Vocalizing in this way will train is to learn to speak to the AI. Because of the humanization of the AI creature, we may come to suspect it of motives which are not present. The design of personalities: the voice of various AI personalities representing various companies or serving different purposes. Chatbot therapist: named Ellie - wired magazine. The question of affordances with AI: James Gibson - there certain things a tech appears able to do so that when we encounter it those functions determine how we negotiate it. The affordances are built into the form: a hammer, a paintbrush. The interface condenses/reduces the potentially infinite functions to make it more manageable for the human user. The design of the personality helps solve the problem Of affordances: certain personalities present themselves as being capable of certain kinds/qualities of tasks.

Always a blurry distinction between programming & interaction: photoshop. Siri is a discontiguous operation, that is: learning to understand your voice, is optimized by the responses that are returned by the user in relation to the info pulled from the internet, & from other users interacting simultaneously - Siri learning to understand what we (humans, hominids) mean when we ask/respond which constitutes an act of programming through interaction. Joseph Weisenbaum: chat bot therapist 1966 - Eriksonian ego, asking in a Socratic dialogue of incessant questions.

Synthetic voice simulator software!! The Markov model: stochastic model, trying to predict what you will say next.

Simulation theory: we live in a simulation.

O'reilly's weekly podcast on bots.

Text/Voice/Image: Derrida reading Platos dialogues: "the metaphysics of presence": text-image-voice sequence. BOT APOPHENIA: a kind of false pattern recognition, a breakdown in correlation-causality: the assumption that one thing causes another = true correlations.

Tay + planet 4Chan = Nazi teen porno

Documentary about Kasparov losing his shit playing chess with IBM deep blue AI: who am I playing? Am I bugged? Have I been duped?

........

Remarks on the w/hole of representation in computer vision.

Diagram <> Interface

Encode cultural bias through the reduction of fidelity in the interface: for function. How do these systems see wrong to allow us to see ourselves differently. What we see as realism it sees as abstraction & vice versa. The equivalent of green face paint architecture. Photosynthesis as a vision without representation. Subjectivity tied to the discourse of the ocular, so that the displacement reorientation the articulations of inside/outside: how are we seen by the machine? It's not without aesthetic judgements: humans as an ambivalent surface vision is projected onto: the clearing away of a closely guarded illusion. We are creeped out by how we ourselves look through the machinists gaze. Productive alienation: Trevor Paglin : photography of machine vision. Cosmists: the communist history was determined by the sun