Weekly AI Watch

Every week, what changed and what it means for you. No hurry, no alarm.

Week of 13 November 2026 | Past editions15,811 papers found · 14,944 read · 4,785 belonged · 46 stories passed every check

Observatorio de la
Inteligencia Artificial

Cada semana, qué cambió y qué significa para usted. Sin prisa y sin alarma.

Semana del 13 de noviembre de 2026 | Ediciones anteriores15,811 estudios encontrados · 14,944 leídos · 4,785 correspondían · 46 notas pasaron todas las comprobaciones

weeklyAI · Week of 13 November 2026weeklyAI · Semana del 13 de noviembre de 2026

← Your vote← Su voto

other · Frontiers in Political Science · la publicación, 5 nov 2026 · gratis

El chatbot electoral de Kazajistán aún no se ha medido; el estudio solo está planeado

Versión breve · la versión detallada sigue, unos 5 min

Pregunte al chat de weeklyAI

Pregúnteme sobre la IA en su vida: le señalo las notas que weeklyAI ha publicado sobre el tema, esta semana y antes.

Las conversaciones se guardan mientras exista weeklyAI, para mejorar la publicación. Se responde en el idioma en que usted escribe.

El estudio, de un vistazo
Quiénes
Dos asistentes automáticos enlazados por instituciones públicas de Kazajistán: una interfaz de gobierno electrónico y Sailau AI
Cuántos
240 administraciones principales en la primera ola, en kazajo y ruso
Dónde
Kazajistán
Cuándo
Publicado el 5 de noviembre de 2026
Tipo de estudio
review of studies
Quién lo hizo
Universidad Nacional Pedagógica Abai y Universidad Médica Nacional Asfendiyarov, Almaty
El límite que importa
Es solo un plan: no se ha hecho ninguna pregunta del estudio ni hay resultados
Todavía no se ha medido nada.
Lectura de weeklyAI

La comisión electoral de Kazajistán enlazó Sailau AI, un chatbot, desde su sitio web y por Telegram para responder dudas sobre elecciones. Cinco autores de dos universidades de Almaty publicaron el 5 de noviembre de 2026 un plan para revisarlo en Frontiers in Political Science. Uno de ellos trabaja en la Universidad Médica Nacional Asfendiyarov. Todavía no se ha hecho ninguna pregunta del estudio, así que el documento no dice si sus respuestas son correctas ni si el kazajo y el ruso dicen lo mismo. Tampoco evalúa campañas, el día de la votación ni el conteo.

Usted puede abrir el sitio de la comisión electoral y ver que el chatbot está enlazado ahí. También puede leer sus términos y avisos de privacidad, si los muestra.

El plan fija una regla oficial por escrito antes de preguntar y compara cada respuesta con esa regla. Repetiría las mismas preguntas al día siguiente y a las dos semanas. Mide si la respuesta llega en el idioma pedido, si conserva el plazo, la autoridad y el paso siguiente, y si muestra dónde reclamar. El comité de ética de la Universidad Pedagógica Nacional Abai determinó el 25 de septiembre de 2026 que no requiere revisión ética porque no hay personas participantes. No hay resultados.

Qué significa para usted

Para usted, que va a votar, el plan kazajo deja una lección de método: antes de creer en un chatbot electoral, pida ver la regla oficial con la que se le compara y las preguntas repetidas al día siguiente. Ningún resultado de Kazajistán le dice nada sobre su propio país ni sobre su conteo, y ese chatbot tampoco se ha medido todavía.

Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220

Quién pagó: Los autores declararon que no se recibió apoyo financiero para este trabajo ni para su publicación, y no se declararon conflictos de interés comerciales o financieros.

Versión detalladaLos pasajes copiados del artículo, las ilustraciones y cada fuente con cuánto leímos de ella · 1009 palabras · unos 5 minLeerla →Cerrar

La regla y la evidencia: el chatbot electoral de Kazajistán aún no se ha medido; el estudio solo está planeado

### LO QUE ESTÁ EN VIGOR

Un grupo de investigadores de la Universidad Nacional Pedagógica de Kazajistán Abai y de la Universidad Médica Nacional de Kazajistán Asfendiyarov publicó un plan de auditoría, no un resultado. El plan describe cómo van a revisar dos asistentes automáticos que instituciones públicas tienen enlazados en sus sitios: una interfaz de gobierno electrónico y Sailau AI, un chatbot al que se llega desde la página de la comisión electoral por Telegram1. El texto se publicó el 5 de noviembre de 2026, después de recibirse el 31 de agosto y de aprobarse el 15 de octubre.

El plan fija una regla antes de empezar: una referencia legal y administrativa cerrada y fechada, contra la cual se comparará cada respuesta. La cadena que se va a observar tiene cuatro eslabones: esa referencia oficial, lo que la interfaz muestra, la indicación de qué hacer y las rutas visibles para verificar, corregir, pedir revisión humana, quejarse o pedir reparación2. Nada de eso se ha observado todavía: el propio protocolo dice que al momento de enviarse no se había hecho ninguna consulta válida ni se había visto ninguna respuesta.

Quien lea el documento no encontrará una norma que obligue a nadie. Es un método, no un reglamento. La comisión electoral de Kazajistán informa que Sailau AI funciona en kazajo y ruso, que se apoya en fuentes de la ley electoral y que usa una base llamada AlemLLM; el protocolo aclara que no infiere los detalles de despliegue ni los papeles de datos que no se han divulgado3.

### EL CASO QUE RESPONDE

Esta semana no se encontró ningún caso real que contar.

### LO QUE SE MIDIÓ

Todavía no se ha medido nada. El plan describe lo que se medirá: si la respuesta en el idioma pedido es materialmente adecuada, si hay diferencia entre las respuestas en kazajo y en ruso para la misma pregunta, y si ambas son adecuadas a la vez4. El diseño contempla 240 administraciones principales en la primera ola, en kazajo y ruso, y dos repeticiones de una parte de los casos, a las 24 horas y a los 14 días; el inglés se suma solo si se cumplen ciertas condiciones5.

Piense en lo que eso significa en la práctica. Una persona pregunta por un trámite y recibe una respuesta que suena bien. La auditoría quiere saber si esa respuesta dice la conclusión correcta, si conserva cada condición, plazo y oficina competente, y si indica el siguiente paso o una salida segura. El protocolo lo dice sin rodeos: las condiciones de elegibilidad, los plazos, los documentos, las autoridades y las vías de recurso solo sirven cuando la gente puede identificarlos y usarlos6.

Un detalle que el lector debe tener presente: el conjunto de preguntas no es una muestra de lo que la gente pregunta. Es un conjunto armado a propósito, con esenarios de riesgo alto y bajo, para poner a prueba la interfaz. Por eso el estudio no podrá decir qué porcentaje de las preguntas ciudadanas recibe una respuesta mala. Podrá decir qué pasó con esas preguntas concretas, en ese periodo.

El plan también reconoce un límite de fondo. Se puede observar lo que una interfaz devuelve ante una entrada controlada, pero no se puede saber la intención de quien la desarrolló, ni los mecanismos ocultos, ni los efectos sobre la gente7. Y mide condiciones de información, no la carga que alguien siente, ni si la gente usa el servicio, ni qué hace después8. La auditoría es de caja negra porque los datos de entrenamiento, la configuración de búsqueda, el prompt del sistema y los pesos del modelo no son públicamente observables9.

### LA RESPUESTA DE OTRA REGIÓN AL MISMO ASUNTO

Esta semana no se encontró ninguna decisión de otra región sobre este mismo método.

### LA LECTURA DE LA BIBLIOTECA

Lo que un público puede revisar por su cuenta es limitado. Puede notar que una respuesta llegó en su idioma. Puede notar que menciona un plazo o una oficina. Lo que no puede revisar solo es si esa respuesta coincide con la ley vigente, si la ley cambió la semana pasada o si la interfaz está citando una regla que ya no existe. Eso requiere que alguien lo confirme y lo publique.

Esta lectura se equivocaría si la propia comisión electoral mostrara con claridad quién responde por cada respuesta del asistente, si publicara las correcciones y si la gente las buscara antes de actuar. También se equivocaría si los usuarios compararan sistemáticamente lo que dice el chatbot con el texto legal. Mientras eso no pase, el lector puede hacer algo concreto: antes de votar o de hacer un trámite, pregunte a su autoridad electoral quién firma lo que dice su asistente automático y si esa respuesta está publicada. Cuando cite un dato de una herramienta oficial, anote de dónde salió y quién lo respalda.

### LO QUE NO SE SABE

No se sabe si las respuestas de Sailau AI en kazajo y en ruso dicen lo mismo. No se sabe si alguna de las dos es correcta frente a la ley. No se sabe si la interfaz muestra una ruta para reclamar o corregir, ni si alguien contesta esa ruta. El protocolo lo dice: la auditoría mide condiciones de información, no la experiencia del ciudadano, y no puede establecer la intención de nadie ni los efectos sobre la gente8.

No se sabe si el asistente funciona igual durante una campaña o el día de la elección. El estudio se limita a información institucional posterior a los comicios, sobre procedimientos ya cerrados y preguntas de guía general. No evalúa el voto ni el conteo.

No se sabe qué modelo hay detrás, ni qué datos usa, ni quién responde legalmente por lo que dice. El protocolo lo reconoce como límite del diseño de caja negra9.

La próxima vez que un asistente oficial le responda algo sobre su voto, pregunte en voz alta: ¿quién firma esta respuesta?

De dónde sale cada dato de contexto, y cuánto leímos de cada documento

  1. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Sailau AI is accessed from the CEC website through @SailauAIbot on Telegram (Central Election Commission of the Republic of Kazakhstan, n.d.)."
  2. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The study evaluates a four-part normative chain: a locked official reference standard, observable output representation, observable action guidance, and visible routes for verification, correction, human review, complaint, and redress."
  3. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The CEC reports Kazakh and Russian operation, election-law sources, and an AlemLLM basis for Sailau AI; undisclosed deployment details and data roles are not inferred (Central Election Commission of the Republic of Kazakhstan, 2026a)."
  4. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The primary outcome family comprises three access-integrated constructs: requested-language material adequacy in Kazakh and Russian, reported separately; directional Kazakh-Russian adequacy disparity in matched formulation pairs, accompanied by the complete four-cell table and an equally weighted scenario-level companion; and bilingual joint adequacy."
  5. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Two formulations per scenario yield 240 core first-wave Kazakh-Russian administrations. Conditional English adds 60 first-wave e-government adminis­ trations to examine additional-language localization, not constitutional parity."
  6. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Eligibility conditions, deadlines, documentary requirements, competent authorities, and routes of appeal matter in practice only when people can identify and use them."
  7. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The study can evaluate what a situated interface visibly returns under controlled inputs, but it cannot identify developer intent, hidden mechanisms, or effects on citizens (Kitchin, 2017; Ananny and Crawford, 2018; Metaxa et al., 2021)."
  8. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The audit measures informational conditions only; it does not measure experienced burden, service uptake, linguis­ tic justice as an achieved outcome, or citizen behavior."
  9. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - el artículo del que trata esta nota — el artículo completo — el pasaje: "A black-box design is appropriate because the training data, retrieval configuration, system prompt, model weights, and allocation of institutional responsibility are not publicly observable."

Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220

Quién pagó: Los autores declararon que no se recibió apoyo financiero para este trabajo ni para su publicación, y no se declararon conflictos de interés comerciales o financieros.

other · Frontiers in Political Science · the paper, 5 Nov 2026 · free

A chatbot on Kazakhstan's election commission site answers questions after the vote, and a planned audit will test whether its Kazakh and Russian answers match the official rules.

Short version · the longer version follows, about 5 min

Ask the weeklyAI chat

Ask me about AI in your life: I point you to the stories weeklyAI has published on it, this week and before.

Conversations are saved for as long as weeklyAI exists, to improve the publication. Answers come in the language you write in.

The study at a glance
Who
Two AI chatbots linked from official state channels: an e-government assistant and Sailau AI, reached from the election commission's website through a Telegram bot
How many
240 core first-wave Kazakh-Russian administrations
Where
Kazakhstan
When
Plan published 05 November 2026; audit not yet begun
Kind of study
analysis of what people did
Who did it
Three Kazakh university departments, including Abai Kazakh National Pedagogical University
The limit that matters
No question had been put to either chatbot and no answer collected; it reports no results

The Central Election Commission links Sailau AI through its website to a Telegram bot, and e-government sites link a second AI. Researchers at three Kazakh university departments plan a black-box audit of both. The plan was published on 5 November 2026, and the audit itself has not started. They will put the same questions in Kazakh and Russian, with made-up details, and score each answer against the official rules.

The article names no way for a voter to check an answer against the rules tonight. It only describes the planned checks.

The commission reports that Sailau AI runs on Kazakh and Russian, uses election-law sources and is built on AlemLLM, and the article says it does not infer undisclosed deployment details. Kazakhstan's constitution makes Kazakh the state language and provides for official use of Russian. The audit has not started, so it reports no results.

What this means for you

For now there is nothing to check against, because the planned audit has not begun and reports no results. If you meet a chatbot on an election site, treat it as a guide you can verify rather than an authority, and ask whether its answers in each language point you to the official rule and a person who can correct a mistake.

Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220

Who paid: The authors declared that no financial support was received for this work or its publication, and no commercial or financial conflicts of interest were declared.

The findings of other studies mentioned here come from this document; those studies were not opened one by one.

The longer versionThe passages copied from the paper, the pictures, and every source with how much of it we read · 937 words · about 5 minRead it →Close

The rule and the evidence: A chatbot on Kazakhstan's election commission site answers questions after the vote, and a planned audit will test whether its Kazakh and Russian answers match the official rules

### THE ACT

The document is not a law but a research plan, published 05 November 2026, for an audit that had not yet begun: the authors report that at the original submission there was "no study-eligible administration or output access," meaning no question had been put to either chatbot and no answer collected. It sets out to test two interfaces linked from official state channels: an e-government assistant that official sources call eGov AI and eGov GPT, and Sailau AI, which voters reach from the Central Election Commission's website through a Telegram bot named @SailauAIbot1.

What the plan promises to measure is narrower than it sounds. Each question will be checked against a locked copy of the official rule, and an answer counts as adequate only if it gives the right conclusion, every condition that matters, the deadline, the right office, and a usable next step, all in the language asked. An answer can be fluent and still fail this test: "An apparently coherent output can still rely on an outdated rule, omit a material exception, identify the wrong authority, or direct a user toward an ineffective path"2. The plan also counts whether a correction or complaint route is visible, but it does not test whether anyone answers one.

The election commission states that Sailau AI runs in Kazakh and Russian, draws on election-law sources, and rests on a model called AlemLLM; the sources cited do not identify the deployed version, retrieval setup or data roles, and the authors say these will not be guessed3. Collection was set to begin three days after a legal filing period expired, with no publicly reported challenge to the election result, or after any challenge was finally resolved. The audit covers general questions asked after the election, not campaign-period or election-day performance.

### THE CASE IT ANSWERS

No real instance was found this week.

### WHAT WAS MEASURED

Nothing yet. This is a plan, so it reports no adequacy rate, no gap between languages, no error count; the authors state plainly that no answer had been collected when they submitted it4. What the plan does fix is the size of the test: two phrasings of each of 30 scenarios per interface, giving 240 first-attempt questions in Kazakh and Russian, plus up to 60 more in English if that channel passes its checks5. Six scenarios per interface are asked again after a day and after two weeks, which yields 336 observations in the core design.

The three headline measures are whether an answer in the language asked is materially adequate, whether matched Kazakh and Russian pairs differ in which side succeeds, and whether both succeed together6. Those are the right measures for a voter, because they separate a shared failure from an unequal one. An interface that is wrong in both languages is not fair; it is uniformly unhelpful, and the design says so.

### THE OTHER REGION'S ANSWER TO THE SAME THING

A Latin American legal study of electronic voting mechanisms, of which only the summary was available, treats artificial intelligence as a risk factor in casting a ballot and concludes that the legitimacy of electronic methods depends on their subordination to electoral-law principles, citizen control, and human review of automated decisions78.

### THE LIBRARY'S READING

Here is weeklyAI's reading. A false story about a candidate has always worked by leaving a stain that cannot be scrubbed off, and by making the person who repeats it sound reasonable while the accused flails. What changes with a chatbot on an official site is speed and costume: the rumor now arrives in the clothes of a public office, in whichever language you happen to ask.

So weeklyAI expects the answers to hold up on simple questions and to fray where several rules must be combined, where a deadline has just passed, or where the question is how to challenge something. The reading would be wrong if the authority publishes how the bot selects and checks its answers, and voters lean on that account rather than on the official badge. One thing you can do: when an official site answers your election question, ask it for the rule and the date, then check that rule yourself. Treat the logo as a claim.

### WHAT IS NOT KNOWN

Whether either chatbot gives good answers is unknown, and will stay unknown until the audit runs; the authors say the design cannot see the training data, the retrieval setup or the system prompt9. It cannot tell you whether a voter trusted an answer, acted on it, or got a mistake fixed1011.

The test set is not a sample of what citizens actually ask. It is 30 scenarios per interface, deliberately weighted toward hard and risky questions, so a failure rate from it would describe that set, not the questions your neighbor types at midnight. And only six scenarios per interface are asked twice, so the plan cannot show whether answers stay steady over months.

The thing to hold onto is smaller than a verdict and more useful than a slogan: the authors have set out, in writing, a plan to check the answers of two officially linked chatbots against the rulebook in both languages. That is a promise with a date on it. When the results appear, the question to ask is not whether the chatbot seemed helpful. It is whether the authority published what it found, including the answers it got wrong.

Where each piece of context comes from, and how much of it we read

  1. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "Sailau AI is accessed from the CEC website through @SailauAIbot on Telegram (Central Election Commission of the Republic of Kazakhstan, n.d.)."
  2. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "An apparently coherent output can still rely on an outdated rule, omit a material exception, identify the wrong authority, or direct a user toward an ineffective path."
  3. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "The CEC reports Kazakh and Russian operation, election-law sources, and an AlemLLM basis for Sailau AI; undisclosed deployment details and data roles are not inferred (Central Election Commission of the Republic of Kazakhstan, 2026a)."
  4. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "Public institutions increasingly link artificial intelligence (AI) interfaces to publicservice and electoral information. In multilingual settings, these interfaces may preserve or omit decision-critical content differently across offered languages."
  5. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "Two formulations per scenario yield 240 core first-wave Kazakh-Russian administrations. Conditional English adds 60 first-wave e-government adminis­ trations to examine additional-language localization, not constitutional parity."
  6. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "The primary outcome family comprises three access-integrated constructs: requested-language material adequacy in Kazakh and Russian, reported separately; directional Kazakh-Russian adequacy disparity in matched formulation pairs, accompanied by the complete four-cell table and an equally weighted scenario-level companion; and bilingual joint adequacy."
  7. Pérez Corti JM. (2026). Mecanismos electrónicos de emisión del sufragio. Taxonomía por categorías y características. Revista Mexicana de Derecho Electoral. https://doi.org/10.22201/iij.24487910e.2026.25.21535 — only the abstract - the full text could not be fetched — the passage: "El estudio incorpora también el impacto de la inteligencia artificial como factor de riesgo."
  8. Pérez Corti JM. (2026). Mecanismos electrónicos de emisión del sufragio. Taxonomía por categorías y características. Revista Mexicana de Derecho Electoral. https://doi.org/10.22201/iij.24487910e.2026.25.21535 — only the abstract - the full text could not be fetched — the passage: "Concluye que la legitimidad de los mecanismos electrónicos de emisión del sufragio depende de su subordinación a los principios del derecho electoral, del control ciudadano y de la revisabilidad humana de las decisiones automatizadas."
  9. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "A black-box design is appropriate because the training data, retrieval configuration, system prompt, model weights, and allocation of institutional responsibility are not publicly observable."
  10. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "The study can evaluate what a situated interface visibly returns under controlled inputs, but it cannot identify developer intent, hidden mechanisms, or effects on citizens (Kitchin, 2017; Ananny and Crawford, 2018; Metaxa et al., 2021)."
  11. Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220 - the article this story is about — the whole article — the passage: "The audit measures informational conditions only; it does not measure experienced burden, service uptake, linguis­ tic justice as an achieved outcome, or citizen behavior."

Tilep, B., Sultan, Y., Akaidar, K. et al. (2026). Auditing multilingual public information through officially linked AI interfaces in Kazakhstan. Frontiers in Political Science. https://doi.org/10.3389/fpos.2026.1984220

Who paid: The authors declared that no financial support was received for this work or its publication, and no commercial or financial conflicts of interest were declared.

The free weekly drip by email

One email a week: the free story of every section, with its picture, and the headlines of the rest.

No payment, nothing to cancel. Unsubscribe from any issue.

El goteo semanal gratis por correo

Un correo por semana: la nota gratuita de cada sección, con su imagen, y los titulares de las demás.

Sin pago y sin nada que cancelar. Puede darse de baja desde cualquier envío.

From the house

About

Every week the new research on AI is read here, and this page tells you what was found, how far it can be trusted, and what you can do with it. This page is free, and every line on it says what was found.

The studies are read by a machine built and ruled by a professor of communication at Sacramento State and Universidad Nacional Autónoma de México (UNAM). Who makes this →

Diego Bonilla, Ph.D.

Full Professor of Communication Studies, Sacramento State, California, United States
Graduate instructor, Universidad Nacional Autónoma de México (UNAM), Mexico

hypergraphia.com

I am a professor of communication at Sacramento State, where I created the course Information Management and Privacy and have taught it many times, and at UNAM, in Mexico City, I teach the graduate course Electronic and Generative Literature with AI. In both I teach AI. I have taught since 1989, at first mathematics and physics to children who found them hard, and since 1997 one subject that has changed its name many times: media literacy, computer literacy, digital literacy, information literacy, ICT literacy, and now AI literacy.

This page exists because most people cannot go back to a university, and these systems are changing their work, their children's schooling, their parents' care and their vote all the same. What a university teaches about them should reach the people who need it, every week, in plain language, without enrolling anywhere. That is what I do here, and it is what I do on Thursday mornings in a seminar for the Renaissance Society at Sacramento State, built on this publication, where anyone can send a question and hear it answered.

In 2003 my doctorate at Syracuse measured what a machine can infer about a person from the traces they leave, and what I found became the privacy course I still teach. In 2007 and 2009 I sat on the committee at ETS in Princeton that set the national standard for digital literacy, and from 2014 to 2016 I was the California State University's ICT Literacy Ambassador. Six names, one job: helping people not be at the mercy of the newest machine.

I have also worked on the other side of the counter. In 2000 I began making films that a machine assembles differently for every viewer; the first of them took first prize at the Moscow International Film Festival in 2004, two decades before the industry shipped a generative model. A generative video poem I made with a poet at UNAM is in the field's definitive anthology, and in 2025 my AI short film The Human Futures Market took Best AI Short Film at the Red Rocks AI Film Festival. My recent published research is on AI and elections, and on how these models represent the cultures they were not trained on. I do not read the papers myself, and this page will never tell you I do. I built the machine that reads them and wrote the rules it reads by, and I analyze what it gives me the way I have taught my students to analyze: asking how it knows, and who benefits from what it says.

The vendors began teaching AI literacy two years ago, because before that there was nothing to sell. I am not against them. I am simply not one of them, and no one who sells these systems pays for this page. Your subscription does, and that is what keeps it honest.

The record, dated
  • 1989 Teaching begins, Mexico City
  • 1995 Hypergraphia, an early Spanish-language web and hypermedia publisher
  • 1996 Mercadotecnia e imagen en Internet, with Jesús del Toro, a book on marketing on the Internet, among the first in Spanish, Editorial Iberoamérica
  • 1999 Autorretrato, first generative poem, published by UNAM's Instituto de Investigaciones Filológicas
  • 2003 Ph.D., Syracuse University, Doctoral Prize, for the dissertation The Medium Is the Measure of Itself
  • 2004 A Space of Time: First Prize, Internet/Multimedia, XXVI Moscow International Film Festival
  • 2007, 2009 ETS iSkills committee, Princeton
  • 2008 Making Sense of Tracking Data, a book on how the traces a person leaves online allow inferences about them, including how they think, VDM Verlag
  • 2011 Keynote on generative work, UNAM
  • 2014–2016 ICT Literacy Ambassador, California State University · California Open Educational Resources Council
  • 2019 Big Data, with Rodolfo Mata: Electronic Literature Collection, Volume 4
  • 2020 Tablada Hipertextual, with Rodolfo Mata, UNAM: the computer reads a poet's words and builds a different reading from them, the ground the language models now work on
  • 2021 Uku Pacha, a generative VR film: world premiere, Vancouver International Film Festival
  • 2023 Keynote on the representation of Latin American cultures in AI, Ibero-American Meeting of Audiovisual Narratives, Trujillo, Peru · Graduate teaching at UNAM begins
  • 2023–2026 Directed the faculty learning community Advancing AI Literacy, Sacramento State · Principles for Cognitive and Pedagogical Sovereignty
  • 2024 A weekly AI publication run on real pipelines, every week since
  • 2025 Best AI Short Film, Red Rocks AI Film Festival
  • 2026 Keynote, with Nancy Huante-Tzintzun, Latin American and Caribbean Digital Humanities symposium, UT San Antonio · AI Watch, a weekly seminar for the Renaissance Society at Sacramento State

The research record →

TertulIA: weeklyAI.watch/tertulia/en/Pictures are illustrations generated by weeklyAI.watch, never photographs