“How do you use artificial intelligence in your work?” This question can now be more important in a job interview than listing individual programs. Companies don't just want to hear if someone can operate a chatbot. They are looking for signs of whether applicants choose tasks sensibly, protect confidential data, verify results, and take responsibility.
This is precisely where the opportunity lies: a convincing answer requires neither a technical profession nor years of AI experience. The key is to explain your own work process concretely. This guide shows how you can present AI in a job interview credibly, which examples work, and which statements tend to trigger distrust.
Why AI competence counts in an interview right now
The AMS JobBarometer describes digitalization as a cross-industry driver and notes that the use of AI applications is gaining importance. This is by no means limited to IT positions. Individual work steps are also changing in assistance, sales, accounting, production, marketing, human resources, or customer service.
In addition, there is a current European framework: Article 4 of the AI Act has obligated providers and operators of AI systems to take measures for the AI competence of their employees since February 2025. According to the current Questions and Answers from the European Commission, regulatory oversight began at the beginning of August 2026. The rule, adjusted in July 2026, does not require a uniform level of training but takes into account role, experience, and context of use. The Austrian AI Service Center of the RTR also emphasizes this context-related approach.
For applicants, this does not mean they have to recite legal knowledge. However, it explains why companies are asking more precisely: responsible handling of AI is increasingly part of professional work capability.
What employers are really testing with an AI question
A question about AI is rarely just a question about tools. Behind it are usually five other questions:
- Task understanding: Do you recognize when AI can make a meaningful contribution?
- Judgment: Do you notice when an answer sounds plausible but is technically incorrect?
- Data protection: Do you know which information does not belong in a freely available system?
- Quality assurance: Can you systematically check sources, numbers, language, and results?
- Responsibility: Do you make the final decision yourself and can you explain it?
Anyone who just says “I work with ChatGPT daily” does not answer any of these questions. A small, comprehensible example is stronger: What was the task? What did the AI take over? What did you check yourself? What measurable result was produced? Where was the tool not allowed to be used?
The five-step answer for AI in a job interview
A good answer takes about 60 to 90 seconds. It can be structured according to five building blocks.
1. Name the task
Start with a real professional, academic, or private project situation. For example: organizing feedback from customer conversations, summarizing a long policy, developing variants for a workshop, or checking a table for inconsistencies. The example should fit the advertised position.
2. Define the AI share
Explain precisely which part was supported. Phrases like “I had initial categories suggested” or “I used the system for three alternative outlines” show more competence than “The AI did that.”
3. Describe verification
Mention your checks: comparison with original sources, spot checks in the table, proofreading by an expert, calculation check, search for omissions, or adaptation to Austrian legal and linguistic customs. This makes it visible that you do not confuse the output with a finished result.
4. Prove the result
Describe the benefit without invented precision. Good evidence includes a faster first draft, fewer overlooked feedback items, a clearer meeting structure, or several tested solution variants. A serious order of magnitude is better than a spectacular percentage without proof.
5. Disclose limitations
Conclude with a limitation: no real customer data in a private account, no unverified legal advice, no automatic personnel decisions, no publication without approval. This restriction does not weaken the answer. It shows professional maturity.
A complete example answer
“In a further education project, I had to evaluate around 80 anonymized feedback items. I used an approved AI tool to suggest initial topic clusters. Afterward, I checked every assignment against the original, merged similar categories, and evaluated critical statements myself. This resulted in a usable structure for the meeting more quickly. I removed personal data beforehand, and the recommendations were only adopted after professional verification.”
This answer contains task, tool role, control, result, and data protection. It appears credible because it neither claims that AI is error-free nor conceals one's own contribution.
How to prepare two robust practical examples
Translate the job advertisement into tasks
Mark three recurring activities of the target role. For an assistant position, these could be appointment preparation, minutes, and information research. In sales, it would be product comparison, conversation preparation, and documentation. Afterward, consider which sub-task AI could support and where human decision-making remains mandatory.
Choose a success and a learning example
Do not just prepare a smooth success. A learning example is often more meaningful: the first output was too general, contained a false assumption, or overlooked Austrian specifics. Explain how you noticed this and improved the process. This shows sensitivity to errors rather than mere tool enthusiasm.
Keep the evidence small and verifiable
A work sample does not need confidential documents. Create a before-and-after example with freely invented data: initial question, short prompt, draft version, your verification steps, and the revised final version. You can also find tips on structured evidence in the jobspot guide Building AI competence on the job.
Which AI competencies become concrete in the CV
The Austrian reference framework DigComp 3.0 AT makes AI-related learning outcomes visible within digital competencies. Europass also recommends describing digital skills with tools, projects, and achievements. For application documents, this results in a simple rule: activity and evidence are more important than a long list of product names.
Instead of “AI: very good,” you can write, for example:
- “Generative AI for structuring anonymized customer feedback; results checked with original data and spot checks”
- “AI-supported research for internal briefings; source verification and final editing under own responsibility”
- “Text and table assistants used for variant generation; no confidential data transferred to external systems”
- “Completed basic course on generative AI and implemented two documented practical exercises”
A short project line simultaneously creates a natural starting point for the conversation.
Data protection: The most important boundary in many answers
The Austrian Data Protection Authority points out that personal data can also be processed in AI systems during the operational phase. Principles such as purpose limitation, data minimization, accuracy, as well as integrity and confidentiality continue to apply.
In the interview, you should therefore not casually mention that you have copied applicant data, customer names, health information, internal contracts, or unpublished business figures into a public AI tool. A secure answer distinguishes between:
- public or self-created practice data,
- anonymized information,
- internally approved systems, and
- data that may not be processed without clear approval.
If you do not know the rules of the potential employer, that is exactly a good follow-up question: “Which AI tools are approved, and which data classes may be processed in them?”
Four difficult questions and credible answers
“Did you write your application with AI?”
Stay honest and separate support from responsibility: “I used AI to check variants for the outline. Experiences, examples, and final formulations come from me; I checked facts and language myself.” The article offers more on clean usage Applying with AI.
“What do you do if the AI provides a wrong answer?”
Name a verification path, not a general declaration of caution: open original source, break down the claim, search for a second reliable source, independently verify the calculation, and mark uncertainty. In case of high risk, approval by a responsible expert follows.
“Which AI tools do you master?”
Name one or two tools actually used and quickly switch to the method. Tool names change. Task analysis, good inputs, quality control, data protection, and documentation remain transferable.
“You have no professional AI experience yet. Why should we trust you with that?”
A small learning project is enough as a start: “I tested three typical tasks of this position with freely available data, documented the results, and noted error classes. Professionally, I would first clarify the internal rules and approvals.” That is stronger than feigned routine.
Three practical cases for different professions
Office and assistance
An applicant has task lists suggested from a fictional meeting protocol. She checks responsibilities and deadlines against the original text and does not automatically adopt anything into the calendar. The proof of competence is not the protocol itself, but the controlled transfer into a workflow.
Craft and technology
An applicant uses AI to draft an understandable explanation for a maintenance step. Technical specifications come exclusively from approved documents; safety decisions and work on the system remain with qualified persons. This shows communication strength without delegating professional responsibility to a system.
Marketing and communication
An applicant develops several subject lines for a practice campaign, defines the target group and tone herself, and checks claims, rights, and brand style. Afterward, she tests variants with a clear metric. The AI provides options, not the strategy.
Warning signs that applicants should avoid
- “I usually adopt the results directly.”
- “With good prompts, the tool makes no mistakes.”
- “Data protection is the IT department's business.”
- Invented time savings or projects that cannot be explained upon inquiry.
- Confidential work samples from a former employer.
- A ten-part tool list without a concrete use case.
- The claim that human control is unnecessary for simple tasks.
What questions applicants can ask the company
A job interview is not a one-way street. With factual follow-up questions, you can find out whether the employer already organizes AI professionally:
- Which AI applications are approved for this role?
- For which tasks is AI explicitly not used?
- How are sensitive data and trade secrets protected?
- Who checks results in case of legal, financial, or personnel consequences?
- What training do new employees receive?
- How are good use cases documented and shared in the team?
These questions fit well into the general preparation for a Job interview in Austria. They show interest in the actual work and not just the next trend.
If no one asks about AI in the interview
Do not force AI into the conversation at any cost. The advertised task remains decisive. A suitable moment arises, for example, when asked about working methods, further training, process improvement, or digital tools. Then a sentence like: “During preparation, I also checked where generative AI could support and what controls would be necessary for that” is sufficient. Afterward, a short practical example follows.
If AI is obviously of no importance for the position, focus on the required skills. An unsolicited long tool presentation can give the impression that you are losing sight of the professional core of the role. Conversely, you may confidently introduce relevant experience if it proves a concrete benefit. The connection to the position should always be audible: not “I am interested in AI,” but “For this recurring task, I could prepare variants faster, but would clarify data approval and professional control first.”
Checklist for the evening before the interview
- Select two AI examples suitable for the position.
- Note task, AI share, verification, result, and limitation for each example.
- Prepare an example of a recognized error.
- Do not take any confidential names, data, or documents with you.
- State tool knowledge only as far as it is practically explainable.
- Formulate a question about approvals, data protection, or training.
- Practice the core answer out loud once in a maximum of 90 seconds.
FAQ on AI in job interviews
Do I have to say unasked that I used AI for the application?
A general disclosure obligation cannot be derived from this. In the case of a direct question, however, you should answer truthfully. It is crucial that all information is correct, your experiences are genuine, and you can stand behind the content yourself.
Do I need an AI certificate?
A reputable course can prove basics but does not replace a practical example. For many roles, a small, documented use case with verification and data protection steps is more meaningful than a certificate without implementation.
Is “Prompt Engineering” a good competence statement?
Only if you can explain what is behind it and what results you have achieved. For many non-technical positions, formulations like “structured task description, iterative improvement, and source verification” are more understandable and credible.
What if the company rejects AI?
Respect the operational rules. Your competence also shows in not using a tool if approval, data protection, or professional suitability are missing. Ask about the reasons and about alternative working methods.
Conclusion: Method beats tool names
Anyone talking about present AI in a job interview should neither feign technical enthusiasm nor dramatize risks. A clear work process is convincing: choose a suitable task, limit the AI share, protect data, check results, and retain responsibility.
Prepare two short examples according to the five building blocks today. If you can explain both without confidential details and also name an error, you already have the most important proof of real AI competence: comprehensible judgment.
