On Micro.Blog @jabel (Jeremy) has been writing about so-called “artificial intelligence” (SCAI for short, my abbreviation) through the lens of Ivan Illich’s Tools for Conviviality. It’s a matter worth taking seriously, and I always appreciate anybody reaching for Ivan Illich, even if I find his work equal parts useful and maddening. For economy, and because it has been awhile since I read Illich, I’ll borrow Jeremy’s quotes from Illich defining conviviality. In a convivial society there is “autonomous and creative intercourse among persons, and intercourse of persons with their environment. … [Conviviality is] individual freedom realized in personal interdependence.” Convivial tools therefore afford people “the freedom to make things among which they can live, to give shape to them according to their own tastes, and to put them to use in caring for and about others.”
By contrast, as Jeremy explains,
the failure of the industrial model of tools is rooted in a key error: namely, that we could make tools that work n behalf of humanity. That, in fact, we could replace human slaves with tool slaves. But we have found that when we replace human slaves with tool slaves, we become enslaved to the tools. Once tools grow beyond their natural scale, they begin shaping their users. The bounds of the possible become defined by the capabilities of the tools.
On one use of SCAI, Jeremy writes:
I think everyone would agree that old-fashioned encyclopedias are convivial tools, i.e., they facilitate autonomous human creativity; they can be picked up and put down at will; they make very few demands upon humans, etc. Search engines, as such, can also be convivial tools in that they are faster, digitized versions of encyclopedias. AI-assisted search might also be convivial in some ways.
The question of whether something like SCAI can be convivial is tempting, but I think it’s a mistake to address it head-on. Instead I want to respond to the first sentence in this paragraph, about “old-fashioned encyclopedias.” In part I want to do this because I am incapable of reading the phrase “I think we can all agree that” without instantly, unconsciously searching for a way to disagree. But in part it may be a useful way of nibbling up to the actual problem of SCAI.
My first thought about encyclopedias is that they have always pretended to be universal—not that they pretend to contain all knowledge, but that they pretend to address a comprehensive range of topics. The word encyclopedia carries etymologically the idea of a closed circle: if a topic is not to be found in a thirty-volume encyclopedia, the implication is that it simply isn’t important. The unspoken premise of an encyclopedia is that experts can, and should, determine the scope of value in knowledge, and I believe that readers absorb that premise unconsciously. That sounds to me precisely like “The bounds of the possible becom[ing] defined by the capabilities of the tools.”
Dictionaries take on similar roles: there are still occasional outcries of protest when this or that word is added to or removed from a given dictionary, because we see the dictionary as bounding and defining the language rather than merely describing it; as giving the language to us. If it were merely a reference, no one would take it personally when some editor tried too hard to be up-to-date. The real problem is whether the people I’m talking with understand what I’m talking about. Language is a tool for conviviality; but is a reference work on language? Given that Illich elsewhere wrote about vernaculars, literally, as human-scale communication—language absorbed rather than language taught—I think I’m on safe ground here saying no.
Of course no reference can be truly, erm, encyclopedic. Any reference has to privilege some sorts of knowledge over others, and so we have to ask what kind of knowledge an encyclopedia privileges.
Inevitably, encyclopedias tend towards universal knowledge of a second kind: that is, knowledge independent of context, that can be applied universally. As a practical matter, most of the knowledge that would be of direct use to me is not universal but local; not context-independent but context-specific. I don’t need to know the history of the seed industry; I need to know which varieties will grow best in the micro-climate of my own yard. I don’t need to know how sugar is produced from cane, I need to figure out dinner for my picky kids. I know, actually, a fair bit about both sugar production and the seed industry; both fascinating in an after-dinner chitchat kind of way but not, you know, actually useful. They have, in my life, a purely aesthetic value, which is not to say they have no value—but it’s the ability to grow and cook food that gives people, quoting Illich again, “the freedom to make things among which they can live, to give shape to them according to their own tastes, and to put them to use in caring for and about others.”
Of course an encyclopedia cannot possibly provide local knowledge, but by its pretense to comprehensiveness it subtly undermines the idea that local knowledge has value. Not that our society doesn’t do enough to undermine local knowledge without the help of an encyclopedia, but still: not, I would argue, convivial.
Related to the matter of local knowledge is that encyclopedias privilege (to borrow an idea from Matthew B. Crawford) knowing that over knowing how. Knowing how to do something used to be the more prized, but today knowing that a thing (however thoroughly abstracted from one’s own existence) is or is not so is the more highly valued. Hence people are thought stupid if they think the sun revolves around the earth but not if they can’t, for example, grow a garden or build a house or perform first aid, even though the ability to feed, house, and care for oneself and others can actually directly improve lives while an understanding of heliocentrism (let alone of general relativity) is, for practically everyone, merely aesthetic. Again, the proliferation of encyclopedias in the 20th century are not to blame for that shift, but they were at least fellow travelers; they aided and abetted it. The way they were used when I was a kid, as the source of first resort for school reports, certainly made clear what kind of knowledge was to be valued: that is, knowledge of the history of Jutland, rather than of how to keep the groundhogs from eating the sweet corn I had so excitedly planted in my back yard. (If you are asking, what or where in hell is Jutland? You might very well ask. You might very well indeed.) I would judge that the knowledge required to use Illich’s “tools for conviviality” (knowledge for conviviality?) is a matter of knowing how, not of knowing that; that the kind of knowledge offered by encyclopedias is of little use (if any) in accomplishing such ends—imagine looking up “first aid” in the encyclopedia while your kid is bleeding profusely!— that, indeed, the values inculcated by the pursuit of encyclopedic knowledge are probably antithetical to one’s becoming a person who values conviviality.
To the inevitable reply that one could replace “encyclopedia” with “reference on how to do things” I would observe that no user manual has ever made me expert at using a tool, and that for all my two or three hundred cookbooks (no really), I cook as well as I do because I’ve done it every day for many years. Now, I do reference them. Reference works can be made to serve convivial ends, surely. But I don’t think that makes them inherently convivial. They may be, depending on the uses of the knowledge they offer and the humility with which they offer it. But I’ve argued in the past that relying on recipes, as linear sets of instructions, not only sidestep skilled work but actually impede its development: one can’t really learn to cook from reading recipes. As reference material for a skilled cook they may assist convivial work, but as tools for education, no — and modern cookbooks generally present themselves as tools for education, not actually for reference.
So maybe the problem is that reference works like dictionaries and encyclopedias are used (and sold) for primary learning rather than for true reference by already-skilled practitioners? Is the same true of internet search engines? Maybe, but it’s hard to see Google making any money (and therefore existing in the first place) by providing only true, literal reference. And while one could say the same of SCAI, that it could be safely useful (and maybe even convivial) to people with sufficient skill to use it wisely — but who are not, of course, the majority of the people who are going to be using it.
Buried in here is another question: what do I mean by inherently? What really are the inherent properties of a tool? Can a tool be good or bad, convivial or destructive, etc., etc. by design rather than by use? As I recall, Illich assumed (but didn’t explicitly argue?) that it could, but this is debatable, and it’s one of those questions that leads so far astray from the really important matter of whether I should use tool x for purpose y that I’d rather find a way around it, if I can. I think the way out is in what I said about the problem of using tools wisely. But that’s a topic for another day.