Christopher Mims reports that some experts working on AI think that the moniker “fuels confusion and hype of the sort that led to past ‘AI winters’ of disappointment” (“Why Artificial Intelligence Isn’t Intelligent,” WSJ, July 31, 2021). I believe the terminology is responsible for the success as well as misperceptions of AI, and that winters are no bad thing.
Mims closes the piece by quoting Melanie Mitchell quoting Herbert Simon saying that the field should be called complex information processing, and has Mitchell musing “What would the world be like if it was called that instead?”
“Complex information processing” is a label only an expert could love. If that had been the name, I strongly doubt that the technologies we describe as AI today would be as pervasive as they are.
“Artificial intelligence” is the language of mythos, and “complex information processing” is the language of logos. Mythos has the mysterious power of a narrative that unifies and energizes a culture at an emotional level. Logos has the ethereal clarity of reason and the aura of objectivity.
Without AI mythos, I doubt we would’ve seen such a massive investment of careers and capital into the field over the decades. AI encapsulates and evokes the godlike dream of creating a living artefact in our own image; making a figure out of clay (silicon!) and breathing life into it. The dream fires people up to do extraordinary things.
However, mythos is not an optional extra. It’s intrinsic to technology since it provides the dreams our stuff is made of (h/t Thomas Disch). Without mythos there wouldn’t be technology, so we have to live with its downsides of confusion and hype. The best we can do is to be prepared for them, or at least not to be too surprised when they happen.
--- 0 ---
Mims notes that words have power and that “names, in particular, carry weight. Especially when they describe systems so complicated that, in their particulars at least, they are beyond the comprehension of most people.” In other words: we have rely on unreliable words to get in a world where we are surrounded by complex things we only vaguely understand.
We should be careful not to overplay the limitation to systems (i.e., to this technology) that are so complicated that they are beyond the comprehension of most people – though, by an implication I’d contest, not beyond the understanding of the experts. First, I would question whether even the experts understand these systems. Second, and contrariwise, I would argue that incomprehension is how we all live our lives, and not just with regard to technology. Even just other human beings are so complicated that each of us is beyond the comprehension of anybody else. But we get by; we’re used to dealing with things that are strictly speaking incomprehensible, often by using words like “intelligence” and “understand” as metaphors. That applies to most artefacts in our lives, from microwaves to cars to Amazon to social media to our governments, let alone to other living things.
--- 0 ---
The dread image of AI Winter is also telling. The intended connotation is that an AI winter is a bad thing. This makes sense in terms of the post-Enlightenment Western ideology of endless progress, especially if investors expect that returns will always be positive. However, in communities embedded in agricultural cycles of life, death and rebirth, winter is a necessary and vital (as in the original Latin vitalis, from vita “life”) part of being. It’s also (arguably) part of mythologies worldwide, see e.g. https://en.wikipedia.org/wiki/Dying-and-rising_god.
Christopher Mims’s article seems to be of the “hype-disappointment” genre. They’re relatively easy to write, since tech lends itself to (is inextricably entangled with?) hype. We’re seeing it at the moment with 5G; I’m also reminded of Solow’s productivity paradox. One of my gripes with the Gartner Hype Cycle is that that it’s actually a one-way trajectory and NOT a cycle! Rather than hype-disappointment, this is just a iteration of a {hype, disappointment} loop.
It’s striking that even articles like this “History of AI Winters” bet on the belief of an upward path; while there are dips, its chart of AI’s popularity is trends ineluctably upwards.
Of course, the image of an AI winter prompts thoughts about an AI spring; however, a search on Google for “AI winter” this morning reported 87,800 results; “AI spring” had only half as many
A Google Ngram Viewer plot for “AI winter” vs. “AI spring” over the period 1970–2019, however, shows “AI spring” with only about a tenth as many occurrences as “AI winter.”
No comments:
Post a Comment