3. Binary computing (computing electronically from a base-2 number system: zeros/off; ones/on) continues and extends our species proclivity to support ourselves with useful objects. It offers tools with which to represent our meanings in text, image, space, object, body, sound, and speech.
3.1. Binary computing, a technology of numbers (zeros and ones) can only count. As such, it computes just four things (and nothing more) from the human repertoire of meaning-making practices:
3.1.1. By ordinal numbering, it takes us beyond the historical confines of natural language by naming the world with an empirical specificity impossible in spontaneous speech: using numbers to create alphanumeric identifiers, barcodes, QR codes, URLs, identifiers in the internet of things, for instance. Many of these have become practically unreadable, which means that objects must now speak their names using the labels that have been applied to them.
3.1.2. It adds things up, at times applying sophisticated statistics and algorithms in the calculation. Such calculations would be too painful and laborious for humans.
3.1.3. It measures things across scales – temperatures, distances and such like, often today with the support of embedded sensors and automated surveillance.
3.1.4. Its peripherals manufacture meaningful things, hence the essential materiality of digital media: written screens or pages, sound recordings, objects in 3D printing, maps of spaces (Cope & Kalantzis, 2019).
3.2. These processes are necessarily reductive, though of course often usefully so. One thing is very complex, not just in itself but the inevitably unique configuration of its context. As soon as you start counting, two or more things thing must be reduced to criterial features. Not only must criterial features be limited; they must also be based on contestable judgments of relevance. Sometimes they are just noise.
3.3. Very useful though these calculating machines have become, they are not intelligence. But they always leverage intelligence – the human intelligence with which we decide what is to be numbered and conclude about the numbers’ meanings. Numbered things must be named, and the names fit into world-theories given to computers. Technically, these are the ontologies that power data tables, database structures and markup. The statistics are meaningless without them. The data does not speak to facts; it speaks to the human meaning frames within which the data have been identified and sorted.
3.4. If we must use the phrase ‘artificial intelligence’ (which – reluctantly – we must because people are talking about it so much nowadays), it can be taken to have one of two, limited meanings:
3.4.1. The equivalent of Alan Turing’s ‘mechanical intelligence’, in the case of binary computing machines that can make calculations which would be too laborious for humans to complete (Turing, 1950); or more narrowly still:
3.4.2. Machine learning, be that supervised (where humans label digital objects such as text and image and the machine looks for similar objects based on their underlying numbering) or unsupervised (where statistical significance or clustering suggests a label might be warranted) (Cope et al., 2021; Zhai & Massung, 2016). In both cases, it is the labels that apply meaning. Computers can’t do this; they can only calculate.
3.5. Minds are irreducibly different from binary calculating machines (Hennig et al., 2021). Only a limited number of the binary propositional logics that can appear in minds (zero/one, if/then) can be used to build electronic computers, and their application is a metaphorical transposition at best.
3.6. The scope of machine intelligence is the scope of calculability, or the transposability of meanings into cardinal (countable) or ordinal (sortable) quantities.
~
COPE, Bill and KALANTZIS, Mary, 2022. The cybernetics of learning. Educational Philosophy and Theory. 6 December 2022. Vol. 54, no. 14, p. 2352–2388. DOI 10.1080/00131857.2022.2033213, p. 2382–2383.