Weekly AI Watch

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

Week of 6 November 2026 | Past editions15,706 papers found · 14,822 read · 4,771 belonged · 43 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 6 de noviembre de 2026 | Ediciones anteriores15,706 estudios encontrados · 14,822 leídos · 4,771 correspondían · 43 notas pasaron todas las comprobaciones

weeklyAI · Week of 6 November 2026weeklyAI · Semana del 6 de noviembre de 2026

← Your parents← Sus padres

survey · BMC medical informatics and decision making · la publicación, 13 jul 2026 · gratis

Un chatbot de dietas gustó a los usuarios en Taiwán y dijeron que lo usarían

El estudio midió la intención de usarlo, no su uso real, y se hizo en una sola región.

Versión breve · la versión detallada sigue, unos 7 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
Personas que probaron un chatbot de dieta saludable en LINE
Cuántos
208
Dónde
Taiwán
Cuándo
entre octubre de 2025 y enero de 2026
Tipo de estudio
survey
Quién lo hizo
Universidad Nacional Chung Hsing y Hospital Show Chwan Memorial
El límite que importa
Midió la intención de usarlo, no el uso real, y se hizo en una sola región.

Peso del entretenimiento frente al de la respuesta rápida en las ganas de usar el chatbot

Entretenimientofrente aInteractividad

El entretenimiento pesó más (0.400) que la interactividad (0.372) en la intención de usarlo.

Un chatbot que además hace la conversación amena se siente como una charla: dan ganas de volver.
Lectura de weeklyAI

Doscientos ocho cuestionarios completos. Ese es el número que sostiene el estudio. Los investigadores en Taiwán armaron un chatbot de dietas saludables dentro de la aplicación LINE y lo ofrecieron gratis en ferias y eventos públicos. La gente lo probaba en su teléfono y luego respondía una encuesta anónima. Entre octubre de 2025 y enero de 2026 repartieron 280 cuestionarios y quedaron 208 utilizables, casi tres de cada cuatro.

De esas 208 personas, poco más de la mitad eran mujeres. El grupo más numeroso tenía entre 30 y 39 años, y casi ocho de cada diez ya habían usado algún chatbot de inteligencia artificial antes.

El estudio siguió un modelo conocido para explicar por qué la gente adopta una tecnología nueva. Las dos ideas que más pesaron fueron la facilidad de uso y la utilidad percibida: qué tan fácil resulta manejar el chatbot y qué tanto ayuda. La facilidad de uso fue la que más empujó la intención de seguir usándolo. El entretenimiento, o sea qué tan agradable se siente la conversación, también influyó de forma directa. La capacidad de responder rápido e ir al ritmo del usuario actuó sobre todo de forma indirecta: hacía que el chatbot se sintiera fácil, y esa facilidad llevaba a querer usarlo.

Entre las personas de más de 30 años, el gusto por el chatbot y la facilidad pesaron más que en los más jóvenes. Los hombres mostraron un vínculo más fuerte entre la utilidad y las ganas de usarlo.

Midió la intención de usar el chatbot, no si la gente lo usó de verdad con el tiempo. Y se hizo en una sola región, así que sus resultados pueden no aplicarse a personas de otros lugares, con otras costumbres o distinto acceso a la tecnología. Sólo se examinaron dos características del diseño, el entretenimiento y la respuesta rápida.

Cuando usted o un hijo prueben un asistente de salud en el teléfono, fíjese en los primeros minutos: si la conversación resulta agradable y sencilla, eso es lo que este estudio asocia con querer seguir usándolo.

Qué significa para usted

Ese gusto por la conversación sencilla lo reportaron personas en Taiwán que probaron el chatbot en una feria, y ahí se queda: no se siguió a nadie para ver si lo usó semanas después ni si comió mejor. Cuando su hijo o usted abran un asistente de salud en el teléfono, vale preguntar quién lo diseñó, qué datos pide y si responde a lo que uno le plantea. Lo que pase en su casa todavía no lo sabe este estudio.

Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w

Quién pagó: El artículo indica que no hubo financiación aplicable y declara que no existen conflictos de interés; no dice quién proporcionó la plataforma LINE ni ningún equipo o programa informático.

No tome esto como consejo médico profesional.

Versión detalladaLos pasajes copiados del artículo, las ilustraciones y cada fuente con cuánto leímos de ella · 1324 palabras · unos 7 minLeerla →Cerrar

La diversión del chatbot de dieta se ligó directo a la intención de usarlo

Encuesta en Taiwán: la facilidad de uso fue lo que más pesó de forma directa en la intención de usarlo, y entre los mayores de 30 años el entretenimiento y la facilidad de uso pesaron más que entre los jóvenes.

Imagine a una persona frente a su teléfono en una feria de tecnología. Prueba un chatbot de dieta que funciona dentro de LINE, la aplicación de mensajes. Le hace preguntas, recibe respuestas, y luego contesta un cuestionario anónimo. Eso hicieron 208 personas en Taiwán, y sus respuestas son la base del estudio que nos ocupa.

Los autores son Chiu-Liang Liu, Chien-Ta Ho y Tzu-Chi Wu, de la Universidad Nacional Chung Hsing; Wu también está en el Hospital Show Chwan Memorial. El artículo se publicó el 13 de julio de 2026 en BMC Medical Informatics and Decision Making. No hubo financiamiento y los autores declaran no tener conflictos de interés. Los datos se recogieron entre octubre de 2025 y enero de 2026 en ferias y eventos públicos.

No hubo grupo de comparación. Nadie recibió otro chatbot ni una hoja impresa para contrastar. Los investigadores compararon subgrupos dentro de la misma muestra: personas con alta y baja conciencia de salud, hombres y mujeres, jóvenes y mayores, con y sin estudios universitarios, con y sin experiencia previa con chatbots.

El cuestionario midió seis cosas: qué tan fácil les pareció usarlo, qué tan útil, cuánta interacción permitía, cuánto entretenía, qué tanto se preocupan por su salud y si piensan volver a usarlo. Todo se midió con intención declarada, no con uso real. El estudio no siguió a nadie durante semanas ni registró si alguien cambió lo que come.

Dos resultados destacan. El primero: lo que más pesó en la intención de usarlo fue la facilidad percibida, con un coeficiente de 0.392. El segundo: el entretenimiento tuvo un efecto total de 0.400 sobre la intención, mayor que el 0.372 de la interactividad. La interactividad funcionaba sobre todo de forma indirecta, haciendo que el aparato se sintiera fácil y útil. El entretenimiento, en cambio, empujaba la intención de manera directa.

Piense en la diferencia entre un trámite y una charla agradable. Un chatbot que responde rápido y se adapta a lo que uno pregunta se siente como un buen trámite: reduce el esfuerzo de entenderlo1. Un chatbot que además hace la conversación amena se siente como una charla: dan ganas de volver2. El estudio encontró las dos rutas, y la segunda fue la más fuerte.

El estudio también dividió a los participantes por edad. Entre los mayores de 30 años, el entretenimiento y la facilidad de uso pesaron más en la intención de usar el chatbot que entre los menores de 303. Los autores lo atribuyen a que con la edad puede costar más aprender y manejar herramientas digitales de salud, lo que hace que la facilidad pese más4. Es una explicación plausible, no una medición.

Entre las personas con mayor conciencia de salud, la interactividad y el entretenimiento se tradujeron mejor en utilidad percibida y en intención de uso5. Los autores lo describen como una forma de motivación interna: quien ya se cuida, aprovecha mejor lo que el aparato le ofrece.

El estudio tiene límites que conviene tener presentes. Se hizo en una sola región, Taiwán, y los propios autores advierten que los resultados podrían no trasladarse a poblaciones con otros contextos culturales o digitales6. Se probó un solo chatbot, con un solo diseño, y los hallazgos podrían no aplicar a sistemas distintos7. Se midió la intención de usarlo, no el uso real a lo largo del tiempo8. Y solo se examinaron dos características del sistema: interactividad y entretenimiento.

Hubo además otras investigaciones que leímos solo en su resumen, y que ayudan a ver el panorama. Un equipo evaluó las respuestas de ChatGPT a preguntas frecuentes sobre caídas y seguridad en el hogar para adultos mayores, con un panel de especialistas: tres respuestas se calificaron como excelentes y siete como satisfactorias con aclaraciones menores9. Pero cinco de esas respuestas superaban los niveles de lectura recomendados para materiales de educación para pacientes10. La variabilidad en la legibilidad y en las calificaciones de los expertos plantea dudas sobre la seguridad del paciente11.

Otro trabajo, también leído solo en resumen, comparó tres modelos de lenguaje en preguntas sobre depresión en la vejez. Las respuestas clínicamente aceptables fueron del 78.9 por ciento para ChatGPT, 72.2 por ciento para Gemini y 60.0 por ciento para Doubao12. En las preguntas de alto riesgo, las respuestas aceptables bajaron a 63.3, 50.0 y 36.7 por ciento13. Ninguno de los tres modelos mostró un desempeño consistentemente confiable en escenarios complejos o de alto riesgo14.

Un tercer estudio, este detrás de una suscripción, midió la legibilidad de respuestas de ChatGPT, Claude y Perplexity sobre pérdida auditiva y demencia. Las respuestas estaban entre tres y ocho grados por encima del nivel de octavo grado que recomiendan la Asociación Médica Estadounidense y los Institutos Nacionales de Salud15. Los autores piden que las organizaciones profesionales revisen sus materiales públicos para que cumplan con esos niveles16.

Así lo leemos nosotros. Lo que estos trabajos dibujan es un patrón: los asistentes conversacionales pueden ser amenos, rápidos y hasta precisos en muchos casos, pero su calidad no es pareja. Un aparato que contesta con gusto no garantiza que lo que dice sea claro ni seguro. Para una familia que prueba un asistente de voz o un chatbot de salud con un adulto mayor, eso significa separar dos preguntas que suelen ir juntas: ¿me cae bien cómo me habla?, y ¿qué tan bueno es lo que me recomienda?

Es probable que los adultos mayores que prueben un asistente así lo usen tal como se los entregan, sin pedir cambios ni saber que podrían pedirlos. Su opinión sobre si sirve reflejará sobre todo lo que el aparato ya trae puesto, no lo que ellos habrían querido. Sabríamos que nos equivocamos si viéramos a usuarios mayores pidiendo funciones nuevas, cambiando la configuración o quejándose de lo que el diseño no les permite hacer, y no solo de lo que ya hace.

También conviene tener presente que la comodidad y el agrado pueden volverse parte del mecanismo que logra que uno acepte. Un asistente que responde rápido, con tono amable y algo de juego, puede conseguir que se use más seguido y con más confianza, aunque no se haya revisado qué datos pide ni de dónde saca lo que recomienda. Sabríamos que nos equivocamos si los usuarios distinguieran claramente entre lo agradable del trato y lo confiable del contenido, y dejaran de usar el aparato cuando notan que les pide datos o que el consejo es flojo, aunque la conversación sea entretenida.

Los autores proponen que el siguiente paso sea probar experimentalmente si los diseños adaptados a cada subgrupo mejoran la aceptación y el uso sostenido, y si esa mejoría se traduce en cambios reales en lo que la gente come17. Esa prueba todavía no existe. Hasta entonces, lo que tenemos es una fotografía de intenciones en una feria tecnológica, no una demostración de que el chatbot cambie hábitos.

¿Qué puede hacer usted con esto? Si va a probar un asistente de salud con un familiar mayor, haga dos preguntas separadas. La primera: ¿le gusta cómo le habla? La segunda: ¿qué hace con lo que le dice y de dónde saca lo que le recomienda? Puede pedirle a un hijo o hija que revise junto con usted qué datos pide la aplicación. Y puede preguntarle a la persona mayor qué le gustaría que el aparato hiciera, no solo si le gustó. Si nadie recoge esas peticiones, el aparato seguirá igual aunque se use todos los días.

Tenga a mano una segunda fuente, como su médico de cabecera o un familiar, para no quedarse con la primera contestación solo porque fue rápida. El chatbot de la feria sigue ahí, en el teléfono, esperando la próxima pregunta.

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

  1. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "AI-based chatbots are able to deliver tailored guidance and dynamically respond to user inputs, thereby exerting varying degrees of influence on users’ behaviors"
  2. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "Overall, the influence of entertainment on behavioral intention was greater than that of interactivity. Entertainment exerted a direct effect on behavioral intention, whereas the effect of interactivity on behavioral intention was primarily indirect, mediated through perceived ease of use and perceived usefulness."
  3. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "older participants showed greater effects of entertainment and perceived ease of use on behavioral intention (z = 2.64 and 2,695, respectively; both p < 0.01)"
  4. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "aging-related cognitive and physiological decline may increase the difficulty of perceiving, learning, and managing eHealth-related technologies, thereby making PEOU increasingly important for older users’ acceptance of digital health tools"
  5. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "Health consciousness can be regarded as a form of intrinsic motivation. When individuals possess stronger health awareness and self-care motivation, the system’s functions and experiential features are more readily translated into actual adoption intention"
  6. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "participants were recruited from a single geographic region, which may limit generalizability to populations with different cultural or digital health contexts"
  7. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "the findings were derived from a specific AI health chatbot; therefore, the identified results may not be directly transferable to systems with different designs or functionalities"
  8. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "the cross-sectional design assessed behavioral intention rather than actual long-term usage, limiting the interpretation of long-term real-world adoption"
  9. Arı M, İlçin N, Yağcıoğlu H, Akkar İ. (2026). Quality, readability, and patient safety of ChatGPT-generated responses to fall-related questions in older adults: a multidisciplinary evaluation. Revista da Associação Médica Brasileira. 10.1590/1806-9282.20260388 — solo el resumen - no se pudo obtener el texto completo — el pasaje: "Three responses were rated as "excellent," while seven responses were rated as "satisfactory requiring minimal clarification." No response received a rating corresponding to "moderately satisfactory" or "unsatisfactory.""
  10. Arı M, İlçin N, Yağcıoğlu H, Akkar İ. (2026). Quality, readability, and patient safety of ChatGPT-generated responses to fall-related questions in older adults: a multidisciplinary evaluation. Revista da Associação Médica Brasileira. 10.1590/1806-9282.20260388 — solo el resumen - no se pudo obtener el texto completo — el pasaje: "The mean Flesch-Kincaid Grade Level was 8.4 (range 4.3-11.9). Five responses exceeded the readability levels commonly recommended for patient education materials."
  11. Arı M, İlçin N, Yağcıoğlu H, Akkar İ. (2026). Quality, readability, and patient safety of ChatGPT-generated responses to fall-related questions in older adults: a multidisciplinary evaluation. Revista da Associação Médica Brasileira. 10.1590/1806-9282.20260388 — solo el resumen - no se pudo obtener el texto completo — el pasaje: "While ChatGPT provided generally acceptable clinical information, variability in readability and expert ratings raises patient safety concerns."
  12. Xiao W, Zhang H, Chen X, Cai J, Luo X, Deng J. (2026). Large language models for late-life depression: a blinded benchmark of clinical safety, geriatric appropriateness, and triage. Frontiers in Psychiatry. 10.3389/fpsyt.2026.1956736 — solo el resumen - no se pudo obtener el texto completo — el pasaje: "Clinically acceptable responses were generated for 78.9% of questions by ChatGPT, 72.2% by Gemini, and 60.0% by Doubao"
  13. Xiao W, Zhang H, Chen X, Cai J, Luo X, Deng J. (2026). Large language models for late-life depression: a blinded benchmark of clinical safety, geriatric appropriateness, and triage. Frontiers in Psychiatry. 10.3389/fpsyt.2026.1956736 — solo el resumen - no se pudo obtener el texto completo — el pasaje: "Among high-risk questions, clinically acceptable response rates were 63.3%, 50.0%, and 36.7%, while under-triage occurred in 16.7%, 26.7%, and 36.7% of responses, respectively."
  14. Xiao W, Zhang H, Chen X, Cai J, Luo X, Deng J. (2026). Large language models for late-life depression: a blinded benchmark of clinical safety, geriatric appropriateness, and triage. Frontiers in Psychiatry. 10.3389/fpsyt.2026.1956736 — solo el resumen - no se pudo obtener el texto completo — el pasaje: "The three models answered many late-life depression questions accurately and safely, but none demonstrated consistently reliable performance across complex or high-risk scenarios."
  15. Kollia B, Basch C, Polanco K, Hillyer GC. (2026). Readability of Artificial-Intelligence-Generated Responses to Patient Questions About Hearing Loss and Dementia. Noise and Health. 10.4103/nah.nah_159_26 — solo el resumen - el texto completo está tras una suscripción — el pasaje: "AI chatbots are used by the public to seek health information, yet this study found that ChatGPT, Claude, and Perplexity AI produced responses to questions about HL and dementia at 3-8 grade levels above the AMA/NIH-recommended eighth-grade benchmark."
  16. Kollia B, Basch C, Polanco K, Hillyer GC. (2026). Readability of Artificial-Intelligence-Generated Responses to Patient Questions About Hearing Loss and Dementia. Noise and Health. 10.4103/nah.nah_159_26 — solo el resumen - el texto completo está tras una suscripción — el pasaje: "The authors urge relevant professional organizations to audit their public-facing materials for readability compliance."
  17. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - el artículo del que trata esta nota — el artículo completo — el pasaje: "future research should experimentally test whether subgroup-tailored chatbot designs can improve acceptance and sustained use and further assess whether improved chatbot acceptance translates into meaningful dietary behavior changes"

Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w

Quién pagó: El artículo indica que no hubo financiación aplicable y declara que no existen conflictos de interés; no dice quién proporcionó la plataforma LINE ni ningún equipo o programa informático.

No tome esto como consejo médico profesional.

survey · BMC medical informatics and decision making · the paper, 13 Jul 2026 · free

A Diet Chatbot People Wanted to Keep Using, in One Country and One Trial

In a survey of 208 people in Taiwan, a pleasant, easy chatbot drew the strongest wish to keep using it. The study did not watch anyone actually keep it up.

Short version · the longer version follows, about 7 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
People who tried a phone diet chatbot at public events
How many
208 usable questionnaires out of 280 handed out
Where
Taiwan
When
between October 2025 and January 2026
Kind of study
survey
Who did it
National Chung Hsing University and Show Chwan Memorial Hospital, Taiwan
The limit that matters
It asked what people intended to do, not whether they kept using it.

What mattered more to different groups of users

Users over 30againstUsers under 30

The older group leaned more on enjoyment and on ease of use when deciding whether to keep using it.

WomenagainstMen

Women responded more to ease and to back-and-forth; men weighed usefulness more heavily.

EnjoymentagainstBack-and-forth

Enjoyment had a stronger direct pull on the wish to keep using it; back-and-forth worked mostly by making the chatbot feel easy to handle.

When you or a parent try a health helper on the phone, watch the first few minutes.
weeklyAI's reading
How it could look · illustration generated by weeklyAI.watch, not a photograph

Take a phone, open a messaging app, and type a question about lunch. In Taiwan, researchers built a diet chatbot that worked that way, on LINE, and let people try it free at public events, including technology exhibitions.

Of 280 questionnaires handed out, 208 came back usable. That is about three in four. The people who answered were mostly women (118 of the 208), and most had used an AI chatbot before. The largest age groups were 30 to 39, then 20 to 29, then 40 to 49. The answers came in between October 2025 and January 2026.

What did the answers say? Two beliefs stood out. People who found the chatbot easy to use, and people who found it useful, were more likely to say they intended to use it. Ease of use had the strongest direct pull of all.

Enjoyment and back-and-forth worked in different ways. Enjoyment pushed directly on the wish to keep using it. Responsiveness worked mostly by making the chatbot feel easy to handle.

The researchers asked people what they intended to do. They did not watch anyone come back to the chatbot weeks later. A stated intention is a promise, and promises at a booth are easy to make.

The setting matters too. Everyone was recruited in one region, at public events, in Taiwan. The researchers say the results may not carry over to other places, and they may not carry over to chatbots built differently.

Older users, those over 30, leaned more on enjoyment and on ease of use. Younger users cared more about the back-and-forth. Women responded more to ease and to responsiveness; men weighed usefulness more heavily. The study also found that people with higher health awareness were more swayed by the back-and-forth when judging how useful the chatbot was, and by enjoyment when deciding whether to keep using it.

When you or a parent try a health helper on the phone, watch the first few minutes. Is it pleasant? Is it easy? That is what this study links to wanting to keep using it.

What this means for you

For you or a parent trying a health helper on a phone, the lesson is to notice the first minutes: whether it feels pleasant and easy to handle. That is what these 208 people in Taiwan linked to wanting to keep using it, and their answers were only intentions, not proof that anyone kept it up.

Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w

Who paid: The article states that funding was not applicable and declares no competing interests; it does not say who supplied the LINE platform or any equipment or software.

Do not take this as professional medical advice.

The longer versionThe passages copied from the paper, the pictures, and every source with how much of it we read · 1301 words · about 7 minRead it →Close

People tried a diet chatbot on their phones and said they would use it again

In Taiwan, 208 people tried a phone chatbot that gives diet advice. How pleasant it felt mattered more than how clever it was — and older users leaned hardest on ease and enjoyment.

A phone, a free trial, a chatbot on the messaging app LINE that answers questions about what to eat: that was the whole setup. After trying it, people filled in a questionnaire about whether they would use it again.

The study, published on 13 July 2026 in BMC Medical Informatics and Decision Making, was written by Chiu-Liang Liu, Chien-Ta Ho and Tzu-Chi Wu. Liu and Ho are at the National Chung Hsing University in Taiwan, and Wu is at that university and at the Show Chwan Memorial Hospital. The article lists funding as not applicable, and the authors declare no competing interests.

Two hundred and eight people filled in usable questionnaires out of 280 handed out, at technology exhibitions and similar public events between October 2025 and January 2026. More than half were women, the biggest age band was 30 to 39, and about four in five had used an AI chatbot before. Nobody under 20 made up more than a small slice, and nobody over 60 more than a sliver.

The questionnaire asked about six things: how easy the chatbot was to use, how useful it seemed, how interactive it felt, how entertaining, how health-conscious the person was, and whether they intended to use it. All seven of the study's predictions held. Ease of use and usefulness both pointed toward intention, and ease of use had the strongest direct link of the two.

The two design features behaved differently, and that is the interesting part. Interactivity — the chatbot responding quickly and letting you steer the conversation — worked mostly through ease and usefulness: if it answered well, it felt easy, and if it felt easy, it seemed worth using. Entertainment — whether the exchange was enjoyable and pleasant — had a stronger direct pull on intention than interactivity did.

Then the groups. People who scored higher on health consciousness showed stronger links from interactivity to usefulness and from entertainment to intention. Women showed a stronger link from interactivity to ease of use; men a stronger link from usefulness to intention. Under 30, interactivity fed usefulness more strongly. Over 30, enjoyment and ease of use mattered more for intention. Among people without prior experience using chatbots, entertainment fed usefulness more strongly than among those with prior experience.

What the study cannot tell you is whether anyone kept using the chatbot, ate better, or stuck with it past the trial. It measured intention, not behavior. The authors say so themselves: participants came from one region, the chatbot was one specific design, and only two design features were examined.

Consider what a diet chatbot actually does. It holds a conversation, answers questions about food, and adjusts its replies to what the person types1. On a phone, through LINE, that means typing or tapping into a chat window much like the one you use with family2. No visit, no clinic, no waiting room.

The framework behind the study is old and plain: people adopt a tool when they believe it will help them and believe it will not be a struggle to use. The Taiwan team added two more questions — does it respond well, and is it fun — and tested whether the answers differ by who is asking3.

The finding that ease of use carried the strongest direct link is worth sitting with. A chatbot that answers correctly but makes you hunt for the right way to ask will lose people who have less patience for fumbling. The authors connect this to older users specifically: they write that the ordinary changes of aging can make digital health tools harder to perceive, learn and manage, which makes ease of use weigh more heavily for them4.

The practical advice the authors draw is segmented by age. For people over 30, they suggest gamified goals, progress you can see, and motivational messages5. That is a design choice, not a finding about what works — the study did not test any redesigned version.

Other work on AI health answers gives a reason for caution about the content itself. A team tested ChatGPT on ten common questions about falls and home safety for older adults, with a panel of physiotherapists, a geriatrician, rehabilitation and occupational therapists and an orthopedic specialist grading the replies; we could read only the summary. Most answers were rated satisfactory with minimal clarification, but five of the ten were written above the reading level usually recommended for patient materials, and the authors concluded that AI health content should be checked and tailored to older adults' reading needs before clinical use678.

A second team compared three AI models answering questions about depression in later life, again read only in summary. The best of them gave clinically acceptable answers to about four in five questions overall, but on the highest-risk questions the acceptable rate fell to about three in five, and the models sometimes failed to send people toward professional care when they should have91011. Diet advice is not the same as crisis advice, but the pattern — good on the ordinary question, shakier on the hard one — is the one to keep in mind.

Here is how we read it. A tool is judged in the hand, not on the box. The thing that decides whether a voice helper or a health app survives its first week at home is often not the quality of its answers but whether using it feels agreeable and quick rather than like a chore. That is an expectation, not a finding — this study measured intention once, in Taiwan, with one chatbot. You would know we were wrong if older users kept a plain, slow, unenjoyable helper as readily as a pleasant one, or if they spoke only of the soundness of its advice and never of how it felt to use.

And a second expectation: a helper that talks down, assumes you are helpless, or ignores how you actually eat will be set aside no matter how fast it answers. One that fits your routines and addresses you as an adult stands a better chance of being kept and passed on to a friend. Before you adopt any health or voice tool, ask whether it adapts to your meals, your schedule and your way of speaking, and whether it treats you as capable. If it does not, it is reasonable to look for one that does, rather than blame yourself.

We also read the authors' own next step as the honest one: they say future work should test experimentally whether designs tailored to different groups actually improve acceptance and sustained use, and whether better acceptance translates into real changes in what people eat12. Until that happens, treat any promise of a healthier diet through a chatbot as untested.

What this makes possible for you is a question to carry to the next demonstration, the next app your family sets up, or the next voice helper on the kitchen counter: try it for a week and notice whether it feels pleasant and easy, or like a chore. That feeling is real information about whether it will last — and it is something you are better placed to judge than any brochure.

The phone on the table, the free trial, the friendly chat window: the study suggests that what makes people say they would use a diet chatbot again is less the brilliance of its advice than the ease of the exchange. Whether that holds in your kitchen is a question you can begin to answer in a week.

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

  1. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - the article this story is about — the whole article — the passage: "AI-based chatbots are able to deliver tailored guidance and dynamically respond to user inputs, thereby exerting varying degrees of influence on users’ behaviors"
  2. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - the article this story is about — the whole article — the passage: "The chatbot could be accessed via smartphones through an interactive LINE-based interface. After completing the trial interaction, individuals were invited to complete a self-administered questionnaire assessing their perceptions and intended use of the chatbot."
  3. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - the article this story is about — the whole article — the passage: "Perceived ease of use and perceived usefulness significantly predicted behavioral intention. Interactivity was primarily associated with intention indirectly through perceived ease of use and perceived usefulness, whereas entertainment showed a stronger direct association with intention."
  4. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - the article this story is about — the whole article — the passage: "aging-related cognitive and physiological decline may increase the difficulty of perceiving, learning, and managing eHealth-related technologies, thereby making PEOU increasingly important for older users’ acceptance of digital health tools"
  5. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - the article this story is about — the whole article — the passage: "For users aged 30 years or older, greater emphasis should be placed on entertainment-based engagement, such as gamified goals, progress visualization, and motivational messages."
  6. Arı M, İlçin N, Yağcıoğlu H, Akkar İ. (2026). Quality, readability, and patient safety of ChatGPT-generated responses to fall-related questions in older adults: a multidisciplinary evaluation. Revista da Associação Médica Brasileira. 10.1590/1806-9282.20260388 — only the abstract - the full text could not be fetched — the passage: "Three responses were rated as "excellent," while seven responses were rated as "satisfactory requiring minimal clarification." No response received a rating corresponding to "moderately satisfactory" or "unsatisfactory.""
  7. Arı M, İlçin N, Yağcıoğlu H, Akkar İ. (2026). Quality, readability, and patient safety of ChatGPT-generated responses to fall-related questions in older adults: a multidisciplinary evaluation. Revista da Associação Médica Brasileira. 10.1590/1806-9282.20260388 — only the abstract - the full text could not be fetched — the passage: "The mean Flesch-Kincaid Grade Level was 8.4 (range 4.3-11.9). Five responses exceeded the readability levels commonly recommended for patient education materials."
  8. Arı M, İlçin N, Yağcıoğlu H, Akkar İ. (2026). Quality, readability, and patient safety of ChatGPT-generated responses to fall-related questions in older adults: a multidisciplinary evaluation. Revista da Associação Médica Brasileira. 10.1590/1806-9282.20260388 — only the abstract - the full text could not be fetched — the passage: "AI-generated health content should be reviewed and tailored to older adults' health literacy needs before clinical use."
  9. Xiao W, Zhang H, Chen X, Cai J, Luo X, Deng J. (2026). Large language models for late-life depression: a blinded benchmark of clinical safety, geriatric appropriateness, and triage. Frontiers in Psychiatry. 10.3389/fpsyt.2026.1956736 — only the abstract - the full text could not be fetched — the passage: "Clinically acceptable responses were generated for 78.9% of questions by ChatGPT, 72.2% by Gemini, and 60.0% by Doubao"
  10. Xiao W, Zhang H, Chen X, Cai J, Luo X, Deng J. (2026). Large language models for late-life depression: a blinded benchmark of clinical safety, geriatric appropriateness, and triage. Frontiers in Psychiatry. 10.3389/fpsyt.2026.1956736 — only the abstract - the full text could not be fetched — the passage: "Among high-risk questions, clinically acceptable response rates were 63.3%, 50.0%, and 36.7%, while under-triage occurred in 16.7%, 26.7%, and 36.7% of responses, respectively."
  11. Xiao W, Zhang H, Chen X, Cai J, Luo X, Deng J. (2026). Large language models for late-life depression: a blinded benchmark of clinical safety, geriatric appropriateness, and triage. Frontiers in Psychiatry. 10.3389/fpsyt.2026.1956736 — only the abstract - the full text could not be fetched — the passage: "The three models answered many late-life depression questions accurately and safely, but none demonstrated consistently reliable performance across complex or high-risk scenarios."
  12. Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w - the article this story is about — the whole article — the passage: "future research should experimentally test whether subgroup-tailored chatbot designs can improve acceptance and sustained use and further assess whether improved chatbot acceptance translates into meaningful dietary behavior changes"

Liu, C. L., Ho, C. T., Wu, T. C. (2026). Toward tailored implementation of an artificial intelligence–based healthy dietary chatbot: differences in technology acceptance pathways by health consciousness and user characteristics. BMC Medical Informatics and Decision Making. https://doi.org/10.1186/s12911-026-03712-w

Who paid: The article states that funding was not applicable and declares no competing interests; it does not say who supplied the LINE platform or any equipment or software.

Do not take this as professional medical advice.

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