ATTARI: A collection of validated questionnaires for measuring AI-related attitudes
On this page, we provide the freely available ATTARI questionnaires, which were developed through a joint effort of psychological researchers at the University of Würzburg (where the project originated), Chemnitz University of Technology, and the Leibniz Institute for Educational Trajectories. Further development was supported by colleagues from the German Institute for Economic Research and Humboldt University Berlin.
Usage guidelines: All ATTARI scales have been published Open Access, supporting their broad and barrier-free application. Nevertheless, all studies using our measures are expected to include a full citation of the corresponding publication (see below). Under this condition, no explicit permission needs to be obtained to use our scales.
However, if you plan to publish a paper focusing exclusively on our measure, or on a modification or translation of it (e.g., validation studies), we kindly ask you to first contact the corresponding authors.
ATTARI-12 — Original scale
In our 2024 publication "Attitudes towards AI: measurement and associations with personality", we introduced the Attitudes towards Artificial Intelligence questionnaire (ATTARI-12), a psychologically grounded 12-item measure designed to assess individuals' general attitudes towards AI technology. Importantly, the ATTARI-12 distinguishes itself from comparable measures by...
- ...capturing the full attitudinal spectrum ranging from very positive to very negative perceptions (i.e., involving both positively and negatively worded items; not just measuring AI-related hopes or fears)
- ...acknowledging the well-established psychological trichotomy of cognitive, affective, and behavioral aspects in human attitudes
- ...allowing its application independent of specific or narrow contexts (as it assesses general attitudes towards AI)
- ...still being an economocial measure that does not take up too much survey time.
Across several studies, we confirmed the high reliability and strong construct validity of the ATTARI-12, and provided clear evidence on its unidimensionality.
When using the ATTARI-12, please cite:
Stein, J.-P., Messingschlager, T., Gnambs, T., Hutmacher, F., & Appel, M. (2024). Attitudes towards AI: Measurement and associations with personality. Scientific Reports, 14, 2909. https://doi.org/10.1038/s41598-024-53335-2
The following expandable tabs provide the instruction, full list of items, and scoring manual in your language of choice.
Instruction: In the following, we are interested in your attitudes towards artificial intelligence (AI). AI can execute tasks that typically require human intelligence. It enables machines to sense, act, learn, and adapt in an autonomous, human-like way. AI may be part of a computer or online platform—but it can also be encountered in various other hardware devices such as robots.
List of items
| Item No. | Wording | Facet | Valence |
|---|---|---|---|
| 1 | AI will make this world a better place. | cognitive | + |
| 2 | I have strong negative emotions about AI. | affective | – |
| 3 | I want to use technologies that rely on AI. | behavioral | + |
| 4 | AI has more disadvantages than advantages. | cognitive | – |
| 5 | I look forward to future AI developments. | affective | + |
| 6 | AI offers solutions to many world problems. | cognitive | + |
| 7 | I prefer technologies that do not feature AI. | behavioral | – |
| 8 | I am afraid of AI. | affective | – |
| 9 | I would rather choose a technology with AI than one without it. | behavioral | + |
| 10 | AI creates problems rather than solving them. | cognitive | – |
| 11 | When I think about AI, I have mostly positive feelings. | affective | + |
| 12 | I would rather avoid technologies that are based on AI. | behavioral | – |
Scoring manual: The ATTARI-12 is usually presented with a five-point answering scale (1 = strongly disagree; 5 = strongly agree). To obtain the final attitude score, first recode all negatively valenced items (no. 2, 4, 7, 8, 10, 12), before averaging all twelve items into a single index. Additionally, three subscores can be calculated for the cognitive, affective, and behavioral component of people's attitudes towards artificial intelligence.
Instruktion: Im Folgenden interessieren wir uns für Ihre Einstellungen gegenüber Künstlicher Intelligenz (KI). Künstliche Intelligenz kann Aufgaben ausführen, die üblicherweise menschliche Intelligenz erfordern. Sie befähigt Maschinen dazu, selbstständig und ähnlich dem Menschen, ihre Umwelt wahrzunehmen, zu handeln, zu lernen und sich anzupassen. Künstliche Intelligenz kann Teil eines Computers oder einer Onlineplattform sein – man kann ihr aber auch in verschiedenen anderen technischen Geräten, wie etwa Robotern, begegnen.
Liste der Items
| Item Nr. | Formulierung | Facette | Valenz |
|---|---|---|---|
| 1 | Künstliche Intelligenz wird die Welt verbessern. | kognitiv | + |
| 2 | Ich habe starke negative Emotionen gegenüber künstlicher Intelligenz. | affektiv | – |
| 3 | Ich möchte Technologien nutzen, die auf künstlicher Intelligenz basieren. | behavioral | + |
| 4 | Künstliche Intelligenz hat mehr Nachteile als Vorteile. | kognitiv | – |
| 5 | Ich freue mich auf zukünftige Entwicklungen im Bereich künstliche Intelligenz. | affektiv | + |
| 6 | Künstliche Intelligenz bietet Lösungen für viele globale Probleme. | kognitiv | + |
| 7 | Ich bevorzuge Technologien, die keine künstliche Intelligenz beinhalten. | behavioral | – |
| 8 | Ich fürchte mich vor künstlicher Intelligenz. | affektiv | – |
| 9 | Ich würde mich eher für eine Technologie mit künstlicher Intelligenz entscheiden als für eine ohne. | behavioral | + |
| 10 | Künstliche Intelligenz verursacht eher Probleme, anstatt sie zu lösen. | kognitiv | – |
| 11 | Wenn ich an künstliche Intelligenz denke, habe ich hauptsächlich positive Gefühle. | affektiv | + |
| 12 | Ich möchte mit Technologien, die auf künstlicher Intelligenz beruhen, lieber nichts zu tun haben. | behavioral | – |
Scoring manual: Der ATTARI-12 wird üblicherweise mit einer fünfstufigen Antwortskala dargeboten (1 = stimme überhaupt nicht zu; 5 = stimme voll und ganz zu). Um den finalen Einstellungswert zu erhalten, codieren Sie zunächst alle Items mit negativer Valenz um (Nr. 2, 4, 7, 8, 10, 12), bevor Sie dann alle zwölf Items per Mittelwertbildung in einen einzelnen Indexwert verrechnen. Zusätzlich können Subscores für die kognitive, affektive und behaviorale Facette gebildet werden.
By now, the ATTARI-12 has been translated into several languages, often involving extensive validation efforts. As these translations have been developed by other research teams, we refer readers to their respective publications.
Please note that the following referrals do not entail an explicit endorsement of scientific rigor or quality.
Since we were not involved in these versions, please direct any further questions to the respective authors.
Korean version: Lee, Kim, & Ahn (2025)
Polish version: Juranek (2025)
Romanian version: Maier et al. (2025)
Chinese version (Simple/Traditional): This translation is currently being developed by Dr. Jian Zhao at the University of Western Australia. We expect it to be published soon.
Japanese version: This translation is currently being developed by Dr. Jian Zhao at the University of Western Australia. We expect it to be published soon.
ATTARI-WHE — Context-specific version focusing on AI in work, healthcare, and education
Complementing the ATTARI-12 (which serves as a measure of context-independent attitudes towards AI), we further developed an applied version of our questionnaire focusing on AI uses in work, healthcare, and educational settings (ATTARI-WHE).
To support its application in time-sensitive large-scale surveys, the ATTARI-WHE comprises only three items per usage context/domain. It was introduced in this 2025 Open Access paper.
When using the ATTARI-WHE in your work, please cite it accordingly:
Gnambs, T., Stein, J.-P., Appel, M., Griese, F., & Zinn, S. (2025). An economical measure of attitudes towards artificial intelligence in work, healthcare, and education (ATTARI-WHE). Computers in Human Behavior: Artificial Humans, 3, 100106. https://doi.org/10.1016/j.chbah.2024.100106
The following table provide the instruction, full list of items, and scoring manual in your language of choice.
Instruction: We would like to know your opinion on artificial intelligence. Artificial intelligence refers to technical devices that can perform tasks that typically require human intelligence. It enables machines to sense, act, and adapt autonomously. Artificial intelligence can be part of a computer program or an online application, but can also be found in various machines such as robots. It can be used in the workplace, in medicine and nursing as well as in education and training.
List of items
| Item No. | Domain | Wording | Facet |
|---|---|---|---|
| 1 | Work | Artificial intelligence offers good solutions for many job tasks. | cognitive |
| 2 | Work | I have a good feeling when I think about the use of artificial intelligence in daily working life. | affective |
| 3 | Work | If I have to complete an important task at work, I would rather choose a technology with artificial intelligence than one without. | behavioral |
| 4 | Healthcare | Artificial intelligence offers good solutions in medicine and nursing. | cognitive |
| 5 | Healthcare | I have a good feeling when I think about how artificial intelligence is being used in healthcare and nursing. | affective |
| 6 | Healthcare | For the treatment of a serious illness, I would rather choose a technology with artificial intelligence than one without. | behavioral |
| 7 | Education | Artificial intelligence is helpful for learning and teaching. | cognitive |
| 8 | Education | I have positive feelings when I think about how artificial intelligence is used in education and training. | affective |
| 9 | Education | If I want to learn something new, I would choose a learning program with artificial intelligence rather than one without. | behavioral |
Scoring manual: The items should be presented in randomized order to avoid sequence effects. Subscale scores and overall scale scores are created by calculating the mean across the three item scores for each domain / facet or the mean across all nine item scores. Scores should only be calculated for respondents with valid responses to at least half of the items.