Antonym: The Light & Shade Edition
Reality isn't black and white
Chiaroscuro (kee-arr-ah-oss-koo-row). You either know the word because it’s part of your world, or you half-remember it from somewhere and almost know what it means. If you feel that second thing – that lack of complete certainty – I want you to resist looking it up. Just notice what it feels like. The itch. The ache of not being sure. You want to be certain, you want to be right, and it’s an itch, an ache that you have to attend to and be free of.
Uncertainty and ignorance. Two of the most uncomfortable feelings we have. No wonder there’s temptation to run to the comfort of polarised tribes that simulate clarity in times of tumult.
I can tell you chiaroscuro is an Italian and you might nod. I can tell you it’s a loanword to English – one of those terms we’ve borrowed wholesale because no English phrase quite does the same job. We use it mainly to describe an artistic technique for making shapes look three-dimensional through contrasts of light and dark.
But chiaroscuro means more than a painting trick. Literally, it’s “light-clear” and “dark-obscure”. Light and shade. Not “black and white” – because that carries connotations of certainty and absoluteness. Chiaroscuro is about gradations. The effects of both at the same time. It can describe a style of illustration or photography, but also the abstract idea of complex realities where things aren’t one thing or the other, but infinitely entangled.
Which brings us to Gramsci.
The Monster in the Chiaroscuro
Antonio Gramsci was a revolutionary and Marxist thinker in Italy, still influential in modern academia and political thought. One of his best-known quotations in English is translated as:
The old world is dying. The new one is slow to appear. And in this interregnum, monsters are born.
Interregnum means the gap between the rule of two regimes. But the word Gramsci actually used wasn’t the Italian equivalent of interregnum. It was chiaroscuro.
Il vecchio mondo sta morendo. Quello nuovo tarda a comparire. E in questo chiaroscuro nascono i mostri.
In the light-clear of the future and the dark-obscure of the past, terrible things appear. Not in a clean break between old and new, but in the murky gradation where both exist at once.
But I digress. (As usual.)
The reason I’m thinking about chiaroscuro is that Anthropic – the company behind Claude – just published what may be the [largest qualitative study ever conducted](https://www.anthropic.com/features/81k-interviews). They - and by they I mean 81,000 instances of Claude – interviewed that many people across 159 countries, in 70 languages, about what they want and what they fear from AI. And the framing the researchers chose for their central finding?
Light and shade.
81,000 People Walk Into a Study
The method here is worth pausing on. This is, by the researchers’ reckoning, the largest qualitative study ever undertaken. I saw someone from one of the big management consultancies noting that a 200-person study of this depth would be hugely expensive. Anthropic did it with 81,000, using Claude itself as the interviewer.
There’s an acknowledged bias in the sample: these are people who already use Claude and who agreed to be interviewed by the AI. But frame it as a study of experienced users rather than the general population and it becomes a lot more interesting. These aren’t attitudes *about* AI from the outside. They’re reflections from people living inside the chiaroscuro already, north of general sentiment, with real experience to draw on.
And they went deep – adaptive, conversational interviews that followed threads into motivations and feelings.
Productivity Is a Feature, Not a Benefit
Before we get to the light and shade, I want to draw your attention to this point, something that puts the AI + productivity obsession back in its corner:
Many others similarly started the interview talking about productivity, but after the interviewer asked about their underlying hope behind it – what realising this vision would enable for them – other priorities surfaced. It wasn’t about doing better work, but increasing their quality of life outside of it. Using AI to automate emails became, in actuality, a desire to spend more time with family.
Fifty per cent of respondents mentioned time savings. But when you pull the thread, time savings isn’t the thing. It’s what the time is *for*. The real aspiration list reads less like a productivity blog and more like a plea: professional excellence (19%), personal transformation (14%), life management (14%), time freedom (11%). People want AI to handle the drudgery so they can be more present, more human, more themselves.
Which is where the light and shade begins.
The Five Tensions
What people want from AI and what they fear from it turn out to be tightly bound. The researchers identified five recurring tensions – directly competing benefits and harms, entangled at the root. They call this the “light and shade” of AI: the same capabilities that lead to benefits also produce harms.
Here they are.
Learning vs. cognitive atrophy. A third of respondents said AI helps them learn – 91% had experienced it directly. But 17% worry about cognitive atrophy: growing so reliant on the tool that you stop thinking for yourself. Teachers and academics reported two to three times higher rates of observed atrophy in students. The jetpack for the mind might also be a crutch.
Better decisions vs. unreliability. The only tension where the negatives outnumber the positives: 37% cite hallucinations and false citations as a real problem. Lawyers, fascinatingly, report the highest rates of both benefit *and* harm. The tool that sharpens your judgement is the same one that confidently fabricates a case citation.
Emotional support vs. dependency. Sixteen per cent describe genuine emotional support from AI; 12% fear dependency. This tension has the strongest entanglement of all five: people who value the emotional support are three times more likely to also fear the dependency. They know. They can feel the pull.
Time savings vs. the accelerating treadmill. Half of all respondents save time. But 18% report that the treadmill just speeds up – expectations rise to fill the gap. Self-employed workers feel both simultaneously: the liberation and the squeeze.
Economic empowerment vs. displacement. Twenty-eight per cent see AI as a path to economic empowerment; 18% fear displacement. This is the most speculative tension, the one least grounded in lived experience so far. The benefits skew towards independent workers (47%) while employees see far less upside (14%). The future is unevenly distributed, as someone once said.
The Itch to Know
The researchers found that people don’t divide neatly into optimists and pessimists. Instead, individuals hold both hope and fear at once. They organise around what they value – and manage the contradictions as they go.
Not black and white. Not “AI is good” or “AI is bad”. Not even the comforting middle ground of “it depends”. Something more uncomfortable: that the light *is* the shade. The capability that helps you learn is the same one that might let your thinking atrophy. The companion that supports you emotionally is the same one you might lean on too hard.
And this connects to something I keep banging on about: AI literacy isn’t a technical skill. It’s a cognitive one. It’s *metacognition* – thinking about thinking. The capacity to notice when the thought-jetpack is carrying you somewhere useful, and when it’s just carrying you. The ability to sit with that uncertainty, that itch, rather than reaching for the false comfort of “AI will save us” or “AI will destroy us”.
Maybe the most important skill we can develop right now isn’t prompt engineering, or workflow optimisation, or any of the technical literacies that fill our feeds. Maybe it’s the capacity to stay in the chiaroscuro. To hold two true things at once. To resist the itch.
The Quote Wall
One more thing. The study includes an [interactive “quote wall”](https://www.anthropic.com/features/81k-interviews) – a filterable collection of voices from across the 159 countries. I’d recommend spending ten minutes with it. The range of people and experiences is extraordinary: farmers, lawyers, students, carers, entrepreneurs, from Lagos to Lisbon to Lima. It’s a reminder that 81,000 is not a statistic. It’s 81,000 people, each sitting in their own particular patch of light and shade.
That’s all for this week.
I’ll be sitting with the contradictions for a while longer – if we don’t want to miss the reality of this strange moment, it’s probably for the best.
Keep thinking. Remember: It’s a great time to have a brain.
Antony
Great summary of an interesting piece of research. Despite the obvious bias in the study (AI users being interviewed by an AI about how they feel about AI) I feel that the outcome describes my personal “mixed feelings”