I am not a fan of the term AI. It is inaccurate and hyperbolic and has troubled me for a long while, but I assumed the thing was lost; everyone just routinely uses the problematic term, and that is how it is going to be. However, I have now reached a point where pedantry seems worth pursuing. I suspect this position could annoy those whose livelihoods are centred around this topic and routinely use ‘AI’, and who carry far more authority than I ever will. Nevertheless, it feels somehow irresponsible to continue using the phrase, especially given my commitment to non-dualism. The brain – and its firing, wiring, neurons – is not a thing that can be popped into a jar, only to carry on working; thinking is feeling and feeling is thinking; meaning, language and material landscapes are highly intra-related aspects of one reality. The implications embedded in ‘AI’ end up reinforcing the brain-in-the jar image, which denies the possibility of a whole range of mechanisms and processes that contribute to critters’ ‘being’ in the world.
Language is always transforming and meaning shifts constantly, but the acceptance of the word ‘artificial’ deepens the false notion that humans and human activity somehow take place outside of nature. By implication, that risks romanticising nature, which can be and often is extremely violent. As for the word intelligence… well, it get’s very complicated indeed.
I have found a useful section in an article titled A high-bias, low-variance introduction to Machine Learning for physicists (Mehta et al., 2019) that explains its use of ML over AI (see Section 18.3). The precision with language throughout is worth noting (although even here, an offending ‘AI’ slips in now and again).
The term AI invokes unrealistic expectations and muddles responsibility. It prompts us to buy into a narrative reminiscent of The Terminator, which risks making us think we have no agency over how these tools can be managed. There is so much hype and doomerism, both of which serve to distract us from addressing the technology with a degree of rationality. The term AI is like a black hole which sucks in all the ire and a good deal of fantasy so we fail to look at the institutions and systems that have long needed overhauling, and out of which the technology in its current form emerged. Somehow, when turning to ML processes, we, as a society, have collectively succumbed to generalisations and groupthink. I expect this has something to do with how complex the processes are; very few of us have any inkling about how it all works, so it is understandable. As Mehta et al. remind us:
This is far from the first time we have seen the use of the term artificial intelligence and the grandiose promises that it implies. In fact, the early 1950’s and 1960’s as well as the early 1980’s saw similar AI bubbles […] These AI bubbles have been followed by what have been dubbed “AI Winters” (McDermott et al., 1985, cited in Mehta et al., 2019)
Another article, more simply titled Why we shouldn’t call Artificial Intelligence ‘Artificial Intelligence’ (a suggestion for a change of nomenclature that is much more important than it appears) (Rodriguez, 2023) is specifically focused on the term, and provides more detailed descriptions of why AI is unhelpful before ending with a list of alternatives. The flaw in the essay is that the author continues to use AI even as he addresses its problems. Nevertheless, it is a concise and useful article that accessibly explains why we need to think about the language we use.
All of this being so, I have just bulk-edited all instances of the terms AI and Artificial Intelligence in the categories and tags on this website so they now read as Machine Learning or ML. ML does not solve things – it continues to be anthropomorphic and also supports a dualist’s view of the human mind and of learning as a disembodied, non-embedded trait – but for the moment, I am following Mehta et al., 2019. If and when I find the right term, I will bulk-edit once more.
Further reading
- An old post in which I refer to The Dawn of Everything: A New History of Humanity (Greaber and Wengrow 2021) and the implications of language.
- Also visit my slow-burning lexicon for thoughts on the words we use today.
Refs:
Mehta, P. et al. (2019) ‘A high-bias, low-variance introduction to Machine Learning for physicists’, A high-bias, low-variance introduction to Machine Learning for physicists, 810, pp. 1–124. Available at: https://doi.org/10.1016/j.physrep.2019.03.001.
Rodriguez, E.Q. (2023) (PDF) Why we shouldn’t call Artificial Intelligence ‘Artificial Intelligence’ (a suggestion for a change of nomenclature that is much more important than it appears), ResearchGate. Available at: https://doi.org/10.13140/RG.2.2.32352.70400.