18 November 2021. AI | Graeber

Robots wrestling with language. Imagining that early people were people is a big step forward.

Welcome to Just Two Things, which I try to publish daily, five days a week. Some links may also appear on my blog from time to time. Links to the main articles are in cross-heads as well as the story.

#1: Robots wrestling with language

The agency Storythings is marking its 10th anniversary by publishing a set of future-themed stories with associated commentaries. I particularly enjoyed one by a Bolivian writer, Edmundo Paz-Soldán, on a robot police officer and their struggles with the slack way in which humans use language.

A couple of extracts from ‘Robot Poet’ will make the point, but it’s reminder that even if we manage to reach that happy state where humans are augmented by robots or AIs, an awful lot might get lost in translation.

(Image by Seanbatty from Pixabay)

And without spoilers, there’s a police investigation going on here, which is why the robot—‘Maturana’—is being talked about in the past tense.

“He had a fixation about language,” you said, “and sometimes he struggled to understand that one thing could mean another: ‘I’ve got a brick in my stomach’ he would say whilst his hands drew a rectangle below his chest. He would spend hours analysing sayings that belonged to what he called popular metaphysics, like ‘one day I slept for three days’, not to mention the verbal mannerisms of some officers, such as ‘a man of many parts he is’.”

The story was originally written in Spanish, so the translator looks to have had some fun here with translating colloquialisms while ensuring they still serve the story:

“Yesterday Maturana asked to meet me and showed me seven hundred printed pages. Notes on linguistic inconsistencies that he had compiled whilst working at the police station. He didn’t understand grammatical duplications such as ‘The exercises were easy-peasy’ or ‘Go and get my coat, go’, why the verb preceded the subject in ‘There stood the table’, why the adverb came first in sentences such ‘As always it hit him’, why the even came at the end in ‘He hadn’t spoken, even.’ He wanted me to explain the rule that explained all these inconsistencies. There must be a rule, he said.

The story also plays off the kind of safeguards that are always talked about when humans and robots mix, but are certain, of course, to go wrong.

The officer telling the story clearly liked the robot. Not so much some of his colleagues—a reminder that some humans will always find ways to score points about status off others.

”He’s got a good heart Gareca used to say, and we kept telling her that that machine hasn’t got one. We saw him kill a robber at the door of a jeweller’s without batting an eyelid.” You didn’t say anything. You recalled that Colque and Mondaca were the very ones who would bombard Maturana with insults and hurtful comments about his inhumanity, and never missed an opportunity to tell him that he was an oddball who pretended to be something that he wasn’t.

It’s an intriguing story, and I recommend it. I didn’t get much from the commentary, which seemed to me to make the story much more literal than it is, but that might just be me.

#2: Imagining that early people were people is a big step forward

I realise that David Graeber has had quite a lot of love here in the course of the past week, but I wanted to come back to his book The Dawn of Everything, co-written with David Wengrow and published after Graeber’s sudden death last year.

The reason for this is that there is an assessment of the book by the academic Justin Smith, who knows this literature well. A couple of points. The first is that he locates Graeber and Wengrow’s work in the other anthropological literature, rather than treating it as if it has arrived from a different academic planet. The second is that he doesn’t spend any time cavilling about their interpretation—if anything, the opposite.

Those of you who follow Smith’s newsletter will know that he writes very long pieces (I don’t know how he finds the time). It also takes him a while to actually get started. So the quotes here represent the briefest of summaries.

A quick reminder about the book: Graeber and Wengrow argue that the classic interpretation on the development of ‘civilization’, proceeding in a more or less linear fashion, from the invention of agriculture to the production of surplus to the invention of private property to stratified societies to the development of the state, is more or less just wrong. Human history, they say, shows much more complexity than this.

Smith notes that Graeber and Wengrow locate the emergence of this classic interpretation in the 1950s and the 1960s:

On the authors’ telling, it is really only in the 1950s and ‘60s, with the quantitatively precise work on daily calorie intake and other such measurables spearheaded by such anthropologists as V. Gordon Childe (1892-1957), that the idea of “man the hunter” took hold, and the default setting of the species was taken to be a seasonally invariant, efficiency-maximizing, and culturally lifeless prehistory. When “man” in “his” “natural” condition is determined to be doing but one thing... the narrative of monolithic unidirectional progress from bands to states becomes vastly easier to maintain.

In their critique of this view, they are building partly on the work on the work of James C Scott, best know for his work on power:

(Scott) has shown that repeatedly and in several different places in human “pre-history”, societies reverted from agriculture back to hunting and foraging, that they did so by choice, and that for several millennia farming existed alongside other viable forms of subsistence in the absence of any well-defined state structure with all its usual indices of inequality. This adaptability should not be at all surprising, given that there are many societies still in existence that alternate seasonally between sedentism and nomadism... But the prevailing view is that there can be no states without sedentism, and that states succeed bands as a “higher” stage of development.

(An indigenous ‘potlatch’ ceremony, Alaska, 1895. Public domain)

But part of the problem here is that we’re not very good at imagining what it was like to live in the deep past. So Smith suggests that maybe the book’s most significant achievement is simply to write about these early modern humans as being, well, people:

Perhaps Graeber and Wengrow’s most affecting accomplishment lies not so much in their new “theory” of the human past, which in any case is only a synthesis of already existing research, as rather in their sympathetic plaidoyer for the singular reality of lives lived in the past, their commitment to the idea that these were real people, as weird and idiosyncratic and unfathomable by quantitative methods as you and I.

'Plaidoyer’? Yes, I had to look it up as well. It’s a plea, or a speech for the defence, borrowed from French.

One of the problems of our societies is that when we think about societies that existed before the industrial revolution, we assume immediately that their lives must have been brutish and short. Thomas Hobbes—writing, of course, in the defence of a state governed by an absolute sovereign—has a lot to answer for.

Smith has a bit of fun with this, at the expense of his mother.

I go to the Whole Foods with my elderly mother, and she looks at the produce section: “Aren’t we lucky to live in this era,” she says, “when there is such a variety and abundance of foods. It really enables us to enjoy it all, rather than just to survive.” This is a sort of entailment from the hypothesis that life before the state was “nasty, poor, brutish, and short”, an existential condition in which, we ordinarily presume, no one had the “luxury” of preferring one food over another, of ever getting their “favorite” for supper.

But we know quite a lot about pre-industrial societies that lived in a state of abundance quite a lot of the time—we just tend of have written them out of the record. In Braiding Sweetgrass Robin Wall Kimmerer spends several chapters on how the indigienous nations of north America ensured abundance. Smith has a similar theme:

“Aren’t we lucky to be here at this festin à tout manger,” every Cree who ever got to participate in an “eat-all feast” is likely to have thought or muttered aloud, each perhaps content with the piece of beaver meat doled out to him in accordance with his social rank, or perhaps aspiring to eat the brains out of the skull some day like the chiefs do, but either way experiencing a condition of abundance and leisure at least as intense as that known by any Whole Foods shopper.

Well, there’s much more of this in Smith’s long, long article, but the point is this. His view on the book is that re-humanises people who have been de-humanised by the narratives of “bourgeois modernity.” You don’t need to know much about the histories of colonialism and capitalism to know that they spend a lot of time de-humanising people, typically before they kill them or exploit them. So one of the strengths of Graeber and Wengrow’s book, on this reading, is that it makes the present strange.

j2t#210

If you are enjoying Just Two Things, please do send it on to a friend or colleague.