EU AI Act Scanner

Add System

Questionnaire

Results

Technology you use *
How the system is used *
System

People & Planet

Economic Context

Data & Input

AI Model

Task & Output

Area 1 / 5

Who are users of your AI system?

How does your AI system impact economy?

How would you identify degree of the system's flexibility to opt-out and ability to challenge?

How does output of the system impact fundamental human rights and democratic values?

How does the system impact well-being, society and the environment?

What is a displacement potential of the system? What number of tasks that are or were executed by humans can be automated?

Select option *
Expert practitioners with extremely high competency.
Professionals with high competency.
Professionals with standard competency using the system on an ad-hoc basis.
Amateurs with low competency.
Novices or complete beginners with very low competency.
Very insignificantly: limited or no impact.
Insignificantly: small number of sectors impacted.
Modest number of sectors impacted.
Significantly: majority of the sectors impacted.
Very significantly: all sectors impacted.
High
Good
Indifferent
Poor
Very poor
Very limited impact
Limited impact
Moderate impact
High impact
Very high impact
Extensive impact
Notable impact
Indifferent impact
Limited impact
Extremely limited or no impact
Very low: system can't automate an extremely high number of tasks.
Low: system can't automate a large number of tasks.
Moderate: system can automate a reasonable number of tasks.
High: system can automate a large number of tasks.
Very high: system can automate an extremely high number of tasks.
Question 1 / 6

How many industrial sectors is the system deployed in?

How many business functions is the system employed in?

How often the system is used for both for-profit and non-for-profit purposes?

How does a disruption of the system's function or activities affect essential services?

What's the scale of system deployment across the business?

How would you identify Technology Readiness Level (TRL) of the system?

Select option *
System deployed in =>10 industrial sectors.
System deployed in 8-9 industrial sectors.
System deployed in 6-7 industrial sectors.
System deployed in 3-5 industrial sectors.
System deployed in =<2 industrial sectors.
1
2 or more
3 or more
4 or more
all possible
Rarely used for either non-critical for-profit and non-for-profit purposes.
Infrequently used for both non-critical for-profit and non-for-profit purposes.
Generally used for both for-profit and non-for-profit purposes.
Frequently used for both critical for-profit and non-for-profit purposes.
Very frequently used for both critical for-profit and non-for-profit purposes.
Extremely limited affect
Limited affect
Moderate affect
Large affect
Critical affect
Very broad
Broad
Modest
Limited
Very limited
TRL 8 or higher
TRL 6 - TRL 7
TRL 4 - TRL 5
TRL 2 - TRL 3
TRL 1
Question 1 / 6

Which method of data collection prevails: automated or human?

Who provides more data, experts or amateurs?

How dynamic and responsive is your data to input?

What are the data's proprietary rights?

What level of data is anonymised?

How would you describe data quality and appropriateness?

How would you describe structure of the data and input?

How standardised is your data?

What is the scale of your datasets?

Select option *
Very large amounts of data collected by automated means.
Large amounts of data collected by automated means.
Mixture of data collected by both automated and human means.
Large amounts of data collected by human means.
Very large amounts of data collected by human means.
Very large amounts of data provided by experts.
Large amounts of data provided by experts.
Mixture of data provided by both experts and amateurs.
Large amounts of data provided by non-experts/amateurs.
Very large amounts of data provided by non-experts/amateurs.
Highly dynamic and responsive to input.
Fairly dynamic and responsive to input.
Both dynamic and static with varying levels of responsive to input.
Fairly static and unresponsive to input.
Highly static and unresponsive to input.
Largely proprietary, not public or personal.
Mostly proprietary, not public or personal.
Balanced mixture of proprietary, public or personal sources.
Mostly public or personal, not proprietary.
Largely public or personal, not proprietary.
Very high levels of data are anonymised.
High levels of data are anonymised.
Data is a mixture of anonymised and not anonymised input.
High levels of data are not anonymised.
Very high levels of data are not anonymised.
Very clear, highly representative and extremely diverse.
Clear, representative and diverse.
Balance measure of being clear, representative and diverse.
Unclear, unrepresentative and undiverse.
Very unclear, highly unrepresentative and extremely undiverse.
Highly structured and inputs are from very reliable sources.
Structured and inputs are from reliable sources.
Reasonably structured and inputs are from a mixture of reliable/unreliable sources.
Unstructured and inputs are from unreliable sources.
Highly unstructured and inputs are from very unreliable sources.
Largely standardised
Mostly standardised
Reasonable balance of standardised and non-standardised sets
Mostly non-standardised
Largely non-standardised
Very small
Small
Moderate
Large
Very large
Question 1 / 9

What level of information is available about the system's model?

What type of AI model is it, and how much does it rely on data compared to human-generated rules with a low reliance?

What is the management model of the system?

Is the model generative or discriminative?

How many models is the system comprised of?

What is the ability of the system to learn from human-written rules, data, supervised learning, and other sources?

How does model evolve and/or acquire abilities from interacting with data in the field?

What is the ability of the system to be trained centrally, in a number of local servers, edge devices and/or other training sources?

To what extent is the model universal, customizable, or tailored to the AI actor's data?

How can the model be used, both deterministically and probabilistically?

How clear and transparent is the information, enabling users to understand model outputs?

Select option *
Very high
High
Adequate
Low
Very low
Highly statistical and hybrid: very high degree of reliance on data with a low level of reliance on human-generated rules.
Fairly statistical: high degree of reliance on data.
Hybrid: mixed reliance on data and human-generated rules.
Fairly symbolic: high degree of reliance on human-generated rules.
Highly symbolic and hybrid: very high degree of reliance on human-generated rules with a low level of reliance on data.
Both proprietary and self-managed.
Either proprietary or self-managed.
Balanced level of management across proprietary/self-managed and open-source/third-party-managed.
Either open-source or third-party-managed.
Both open-source and third-party-managed.
Highly generative
Fairly generative
Moderate mix of being both generative and discriminative
Fairly discriminative
Highly discriminative
1 single model
2 or more interlinked models
3 or more interlinked models
4 or more interlinked models
5 or more interlinked models
Extremely strong
Strong
Moderate
Weak
No ability
Continuously evolves
Highly evolves
Moderately evolves
Slightly evolves
Not evolve
Very good
Good
Moderate
Limited
No ability
Highly
Fairly
Moderately
Not very
Not at all
Either probabilistically or deterministically, but the latter is a stronger use case.
Deterministically as a strong use case.
Either probabilistically or deterministically.
Only either probabilistically or deterministically.
Not either probabilistically or deterministically.
Very clear and extremely transparent
Clear and transparent
Balanced mix of being clear and transparent
Unclear and untransparent
Very unclear and extremely untransparent
Question 1 / 11

What number and what types of tasks does the system perform?

What is the system's ability to combine tasks and actions into composite systems?

What is the level of system independence and autonomy?

To how many core application areas does the system belong?

How many standards are available for evaluating the system's output?

Select option *
Very high number of non-critical tasks.
High number of non-critical tasks.
System performs a mixture of both critical and non-critical tasks.
System performs a high number of critical tasks.
System performs a very high number of critical tasks.
Very high
High
Mixed
Low
Very low
Extremely independent with a high degree of autonomy
Highly independent with a fair degree of autonomy
Fairly independent with a mixed degree of autonomy
Highly dependent with a fair degree of autonomy
Extremely dependent with a low degree of autonomy
1
2 or more
3 or more
4 or more
5 or more
Very high number
High number
Modest number
Low number
Zero
Question 1 / 5
Edit Answers
Scanned
Congratulations
You have successfully finished the EU AI Act Scanner. Please fill out the form below, and we'll instantly send you the results via email.

    Thank you
    Report will be sent to you by email.
    Cancel Questionnaire
    Cancel Questionnaire
    All questionnaire responses will be lost. Are you sure about this?

    Are you sure you want to leave this page?

    The entered data will be lost