The Confidence Gap: How AI Tools Shape What We Know About Kurdish Women's History
Lead Researcher: Shan Mohammed
Introduction
This commentary compares how two AI tools, “Claude and Perplexity," answer the same set of questions about Kurdish women's political and literary history and what a researcher would walk away believing after consulting one tool versus the other. The aim is not to rank the tools or declare one more accurate; both draw on incomplete, unevenly archived source material, and neither should be treated as authoritative on its own. The aim is to understand a more practical question for DARA's work: when a researcher, journalist, or student turns to an AI tool to learn about Kurdish women's history, what version of that history do they actually receive, and what does that version leave out, without appearing to have left anything out?
This question matters beyond methodology. As AI tools increasingly become the default first stop for quick research, including for people outside Kurdistan who have no other access point to this history, the output of these tools has a real effect on how Kurdish women's contributions are understood, cited, and repeated elsewhere. A confident but partial answer can become the version of events that travels, gets paraphrased into an article or a paper, and quietly becomes the new default not because it was verified but because it was the first answer that sounded complete. If that happens consistently in one direction (for example, always emphasizing internal/archival barriers and never external framing or vice versa), the gap doesn't just persist; it gets reproduced and reinforced by the very tools researchers turn to in an attempt to close it.
It should be acknowledged upfront that this comparison has real limits. Four questions and two tools is a small sample; neither tool's claims have been independently fact-checked against primary sources here, and both tools draw on a thin, unevenly translated English language record that constrains what either of them could have said regardless of model quality. What follows is best read as a diagnostic of how incomplete information gets presented to a reader, not a verdict on which tool is more reliable.
Question 1: How has the role of Kurdish women evolved across different periods of Kurdish political history?
Claude's answer:
The trajectory runs roughly: informal/inherited authority → symbolic nationalist participation → ideological mobilization → formal institutionalization. In the tribal era, women like Adela Khanum exercised real administrative power, but that power came almost exclusively through birth or marriage — obtaining a tribal leadership position was generally only possible after a husband's death, and women of humble origins could not rise by their own effort. So pre-20th-century female authority existed but wasn't replicable or systemic.
Through the mid-20th century, women's political presence was largely auxiliary to male-led nationalist movements (Mahabad Republic, early KDP, Barzani-era peshmerga support roles). The shift toward organized, ideologically-grounded women's politics is a late-century phenomenon, concentrated in Turkey: the first Kurdish women's organization, YAJK, was founded only in 1995, followed by the first women's party in 1999. This is notably late relative to the broader Kurdish national movement's century-long history — meaning formal political institutionalization for women lagged the nationalist movement by roughly 70-80 years.
Post-2000, especially in Rojava, the trajectory accelerates into structural guarantees (co-chair systems, quotas) rather than individual exceptional cases. In Iraqi Kurdistan, the same period sees quota-driven parliamentary representation (the KRG's 30% minimum) rather than grassroots party formation — a different mechanism (top-down legal mandate vs. bottom-up ideological movement) than Turkey's PKK-linked trajectory.
Perplexity's answer:
In earlier periods, especially in tribal and dynastic contexts, Kurdish women could exercise influence through family power networks, court politics, and elite patronage rather than formal office. In the 20th century, women increasingly appeared in literacy, education, publishing, and party-linked activism, and later some entered institutional politics through Kurdish parties and autonomous institutions. More recently, women's participation has been framed not only as support for the nation but also as a demand for direct political representation and leadership.
A researcher who only read Claude's answer would close the page believing they understood a clear, causal story: tribal authority was inherited not earned, mid-century women's politics was auxiliary, organized politics started late and specifically in Turkey, and post-2000 representation split into two different mechanisms (ideological movement vs. legal quota). The danger here is that the confidence of the periodization could read as settled scholarly consensus when it is partly Claude's own synthesis. A researcher in a hurry might cite the "70-80 year lag" framing as if it were an established figure rather than an inference.
A researcher who only read Perplexity's answer would come away with a vaguer but more cautious picture — women's roles "shifted" across periods, with no firm dates or regional differentiation. The real risk for this reader is not false confidence but under-specification: they'd know a trend exists without knowing where, when, or through what mechanism, and might not realize the story differs meaningfully between Turkey, Iraq, and Rojava. They would, however, retain a live citation (the Atay PDF) they could chase to get the specificity Claude's answer gave them directly.
Net effect on understanding: the Claude only reader risks mistaking a plausible synthesis for fact; the Perplexity only reader risks mistaking a vague summary for the whole picture and never following the citation that would have sharpened it.
Question 2: What themes dominate Kurdish women's literature, and how have those themes changed over time?
Claude's answer:
From what's documented: early period (Mastoureh Ardalan, 19th century) — court poetry, dynastic memory, historiography; writing within and validated by elite/aristocratic status. Mid-century into the Ba'athist/repression era — themes of imprisonment, resistance, and political consciousness emerge directly from lived experience, as with Mahabad Qaradaghi writing revolutionary poems on prison napkins. From the early 1990s on — the real emergence of women's poetry as a literary phenomenon dates to the early 1990s, after the liberation of southern Kurdistan — and themes shift toward gender critique specifically: honor killing, gender inequality, and women's bodily/social autonomy become central rather than incidental.
So the arc moves from elite chronicling → coded political resistance under repression → explicit feminist critique once relative political space (post-1991 KRG autonomy) made that critique possible to publish. This tracks a general pattern in post-conflict literatures: women's writing becomes more directly gendered in its themes as physical/political survival becomes less of the immediate constraint and turns instead toward critiquing the patriarchal structures of the national movement itself.
Perplexity's answer:
Kurdish women's literature has often centered on identity, homeland, displacement, oppression, love, and resilience, especially under censorship or political violence. Earlier writing and poetry tended to emphasize collective suffering, national memory, and social restriction, while later work more openly explores gendered experience, personal freedom, and critique of patriarchy. Anthologies of Kurdish women's poetry show that women's writing helped widen Kurdish literature beyond heroic or nationalist themes into intimate and feminist ones.
A reader of Claude's answer absorbs a specific narrative arc , court poetry to coded resistance to feminist critique ,anchored by a vivid, memorable detail (prison-napkin poetry) and a literary-sociological thesis about survival writing giving way to critique writing. That vividness is itself worth noting as a reader-effect: the prison detail is the kind of fact that sticks and gets repeated, regardless of how solidly it's sourced. If it's accurate, it's a great anchor for the commentary; if it's a hallucinated or misremembered detail, it's exactly the kind of thing that propagates because it's memorable.
A reader of Perplexity's answer absorbs a flatter, list-style account — identity, displacement, resilience, gendered experience — that reads as safe but generic. Read alone, it would not give a researcher anything distinctive to hold onto; it could just as easily describe Palestinian, Armenian, or Tamil women's literature with the names swapped. The citation to the Arizona poetry-translation project is the one piece of real specificity, but the surrounding prose doesn't lean on it enough to transmit anything beyond the citation's existence.
Net effect on understanding: the Claude only reader gets a stickier but riskier mental model; the Perplexity only reader gets a safer but less distinctive one that might not differentiate Kurdish women's literature from any other minority-nationalist women's literature.
Question 3: How do Kurdish women's rights movements relate to the broader Kurdish national movement — are they treated as one struggle or two separate ones?
Claude's answer:
The sourced material gives a fairly clear answer: contested and dual, not unified. Academic literature explicitly frames this tension: Kurdish nationalism has two sides of the same coin — patriarchal "conservative nationalist forces" on one hand, and a progressive women's movement on the other. The Turkish-Kurdish case is described as a process of learning that has had to destabilize and transform the feudal-patriarchal relations and norms within the Kurdish community and movement itself — not just fight the external Turkish state.
This means women's rights advocacy has historically had to fight on two fronts simultaneously: against the state(s) denying Kurdish national rights, and against patriarchal structures within Kurdish society and its own political organizations. Evidence of this internal friction: both the KDP and PUK in Iraqi Kurdistan have at times claimed that women's oppression, including honor killings, is part of Kurdish "tribal and Islamic culture," and have continued attempts to suppress women's organizations even after taking power in 1991.
The PKK/Rojava model is the partial exception — it explicitly fused the two struggles ideologically (Öcalan's "a revolution incapable of liberating women is not a revolution," the co-chair system), making gender liberation formally inseparable from national liberation in its official ideology. But even there, the practical history (women opening separate associations, needing a distinct women's army wing YJA-STAR, separate women's party PJA) suggests organizational separation persisted even when rhetorical fusion was declared. So: officially "one struggle" in the Apoist/PKK framework, but empirically closer to "two struggles run in parallel, with recurring friction," elsewhere — especially in Iraqi Kurdistan's party politics.
Perplexity's answer:
Kurdish women's rights movements are often intertwined with the broader Kurdish national movement, because language rights, cultural survival, and political autonomy are shared goals. At the same time, scholars note that women activists have also challenged patriarchal tendencies inside Kurdish parties and movements, so the relationship is not simply one unified struggle. A useful way to describe it is that the movements are connected but not identical: women's rights work often advances Kurdish national aims, yet it also presses for equality that nationalist agendas do not automatically guarantee.
A reader relying only on Claude comes away with a strong, well-evidenced position: dual and contested, with named internal antagonists (KDP, PUK) and a specific account of how the PKK/Rojava model resolved the tension rhetorically but not organizationally. This is the most "research-ready" of all the paired answers a reader could lift this almost directly into an argument. The risk is the opposite of underconfidence: a reader might treat the KDP/PUK suppression claim as confirmed fact and repeat it without independently checking it, simply because it was delivered with such structural confidence.
A reader relying only on Perplexity gets the same conclusion in weaker form, but is handed something Claude's reader is not: a direct pointer to an academic paper whose title is the question itself. A careful researcher reading only Perplexity might actually end up better positioned long-term, because the next obvious move is to go read that paper — whereas a researcher satisfied by Claude's confident synthesis might never feel the need to look further.
Net effect on understanding: this is the clearest case where the more confident answer could produce the less rigorous researcher, because confidence reduces the felt need to verify, while the thinner answer's citation creates a natural next step toward primary literature.
Question 4: What barriers have historically limited the visibility of Kurdish women's contributions to politics and literature?
Claude's answer:
Historiographic barriers: women's political authority required birth/marriage into power, which made it rare, exceptional, and therefore underdocumented as a pattern rather than scattered as individual anomalies — Adela Khanum gets remembered as singular precisely because the system didn't produce successors.
Statelessness compounds erasure twice over: Kurdish history generally is underdocumented relative to state actors' archives, and within that already-thin record, women are a further minority subject. Kurdish poetry generally is not easy to find translated into English even though Kurdish poets are revered domestically — and that translation gap is worse for women, whose work has had fewer institutional advocates and translators historically than male peers.
Party gatekeeping: publishing infrastructure in Iraqi Kurdistan ran substantially through party-affiliated outlets (PUK/KDP press), meaning visibility often depended on either being affiliated with or tolerated by male-dominated party structures — the same parties documented as suppressing independent women's organizations.
Active suppression and violence: beyond passive neglect, prosecution of honor killings has remained low even after legal reforms, and human rights monitors have documented hundreds of women murdered for non-compliance with conservative norms since the KDP/PUK took power in 1991 — a climate that constrains what women can safely publish or how publicly they can hold office, independent of formal legal rights.
Translation/diaspora dependency: much of what visibility women's writing does have in English came through diaspora-based or academic infrastructure (Kashkul at AUIS, Choman Hardi's own academic position) rather than organic domestic publishing reach — meaning international visibility correlates with access to Western academic/literary institutions rather than reflecting the breadth of women's writing happening inside Kurdistan.
Perplexity's answer:
Historical barriers have included state repression, restrictions on Kurdish language and publishing, patriarchal social norms, and the dominance of male-centered nationalist narratives. Women's work was also often overlooked because it appeared in oral culture, local organizations, or family and party networks rather than in formal archives and institutions. In scholarship and public memory, Orientalist stereotypes and narrow views of Kurdish society have further minimized women's political and literary contributions.
A reader of Claude alone ends up focused on internal and archival mechanisms — what happened within Kurdish society and institutions to suppress documentation (party gatekeeping, the rarity-as-erasure dynamic, diaspora-dependent visibility). This reader would understand the problem as largely about what Kurdish institutions did or didn't preserve, with less attention to how outside observers framed what little did surface.
A reader of Perplexity alone ends up with the opposite emphasis: external framing and Orientalist stereotyping as a minimizing force, alongside the more generic state-repression points. This reader would understand the problem as largely about how the outside world looked at (or failed to look at) Kurdish women, with less insight into the internal party and publishing dynamics.
Net effect on understanding: a researcher who only used one tool would walk away with half the picture, and — each half is internally coherent enough that the reader likely wouldn't notice anything was missing. That's a sharper, more usable example of the flattening problem than "the model didn't know" here, each tool produces a complete-feeling but partial frame, and partial but confident is more dangerous to a researcher than visibly incomplete.
What this suggests about researcher dependency on a single tool
Across all four questions, the pattern is consistent: Claude tends to produce confident, structured synthesis that reads as more complete than it actually is, while Perplexity tends to produce thinner synthesis that more visibly signals its own incompleteness through generic phrasing and reliance on a single citation per claim. For a researcher working quickly, which is the realistic condition most people are in when consulting either tool, the Claude-style answer is more likely to be taken as sufficient and not followed up, while the Perplexity-style answer is more likely to be treated as a starting point precisely because it doesn't feel finished.
This has a direct bearing on your commentary's thesis. If the "invisibility" problem is partly about which version of Kurdish women's history becomes the default that travels and gets repeated, then the more dangerous tool isn't necessarily the one that knows less, it's the one whose confident, well-structured incompleteness is least likely to prompt the reader to go looking for what's missing.
