{ "$id": "https://standaard.algoritmeregister.org/schemas/fraia.schema.json", "$schema": "https://json-schema.org/draft/2020-12/schema", "title": "Impact Assessment Fundamental rights and algorithms", "description": "", "type": "object", "properties": { "name": { "name": "Name", "category": "BASIC INFORMATION", "type": "string", "description": "What is the colloquial name used to identify this algorithm? Example: 'WMO prediction Rotterdam' or 'Crowd-monitoring Enschede'.", "required": true }, "organization": { "name": "Organization", "category": "BASIC INFORMATION", "type": "string", "description": "What is the full name of the organization used to identify responsibility for use of the model, algorithm or AI? Example: 'City of Amsterdam' or 'Water Authority Limburg '.", "required": true }, "department": { "name": "Department", "category": "BASIC INFORMATION", "type": "string", "description": "What is the full name of the department or division used to specify responsibility for use of the model, algorithm or AI? Example: 'District Segbroek' or 'Traffic and transport'.", "required": true }, "description_short": { "name": "Short description", "category": "BASIC INFORMATION", "type": "string", "description": "Please give a short description of maximum 150 chars in order to give a quick overview of the purpose of the model, algorithm or AI. Example: 'The traffic light priority algorithm prioritises traffic modalities based on applicable law and local regulations'.", "required": true }, "status": { "name": "Status", "category": "BASIC INFORMATION", "type": "string", "description": "Please indicate wether the model, algorithm or AI is in development, in use, or archived.", "required": false }, "revision_date": { "name": "Revision date", "category": "BASIC INFORMATION", "type": "string", "description": "This is the date before which this entry has to be revisited.", "required": true } , "1.1.1": { "name": "Reason and problem definition", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "Explain your proposal for the use/deployment of an algorithm. For which problem is the algorithm to provide a solution? What is the actual occasion or reason to use an algorithm? Why does it require an algorithm?", "required": false } , "1.2.1": { "name": "Objective", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "What is the objective that the use of the algorithm needs to achieve? What is the main objective here, and what are secondary objectives?", "required": false } , "1.3.1": { "name": "Public values", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "What are the public values that prompt the use of an algorithm? If there are several public values prompting the use of an algorithm, can they be ranked?", "required": false }, "1.3.2": { "name": "Public values", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "What are the public values that may suffer as a result of using an algorithm?", "required": false } , "1.4.1": { "name": "Legal basis", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "What is the legal basis of the use of the algorithm and of the targeted decisions that will be made on the basis of the algorithm?", "required": false } , "1.5.1": { "name": "Stakeholders and responsibilities", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "Which parties and persons are involved in the development/use/maintenance of the algorithm?", "required": false }, "1.5.2": { "name": "Stakeholders and responsibilities", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "Have the responsibilities with respect to the development and the use of the algorithm been transparently allocated? How has it been safeguarded that these responsibilities continue to be transparently allocated once the development of the algorithm has been completed and it is being put to use?", "required": false }, "1.5.3": { "name": "Stakeholders and responsibilities", "category": "Intended effects – objectives – preconditions", "type": "string", "description": "Who is ultimately responsible for the algorithm?", "required": false } , "2A.1.1": { "name": "Assessment: Algorithm type", "category": "Data – input", "type": "string", "description": "Is it already (roughly) known what type of algorithm will be used?", "required": false } , "2A.2.1": { "name": "Data sources and quality", "category": "Data – input", "type": "string", "description": "What type of data is going to be used as input for the algorithm and from which sources has the data been taken? If no input data is used, proceed to topic 2A.4.", "required": false }, "2A.2.2": { "name": "Data sources and quality", "category": "Data – input", "type": "string", "description": "Is the quality and reliability of the data sufficient for the intended data application? Please explain.", "required": false } , "2A.3.1": { "name": "Bias/assumptions in the data", "category": "Data – input", "type": "string", "description": "What assumptions and biases are embedded in the data? How are their influences on the algorithm's output corrected or otherwise overcome or mitigated (see also annex 2)?", "required": false }, "2A.3.2": { "name": "Bias/assumptions in the data", "category": "Data – input", "type": "string", "description": "If training data is used: is the data representative for the context in which the algorithm will be used?", "required": false } , "2A.4.1": { "name": "Security", "category": "Data – input", "type": "string", "description": "Is the data sufficiently secure? Differentiate between input data and output data.", "required": false }, "2A.4.2": { "name": "Security", "category": "Data – input", "type": "string", "description": "Is access to the data supervised?", "required": false }, "2A.4.3": { "name": "Security", "category": "Data – input", "type": "string", "description": "Is there compliance with relevant rules on archiving?", "required": false } , "2B.1.1": { "name": "Algorithm type", "category": "Algorithm – throughput", "type": "string", "description": "What type of algorithm will be used? Differentiate between: a. A non-self-learning algorithm in which humans specify the regulations the computer must observe; b. A self-learning algorithm, in which the machine itself is finding patterns in the data.", "required": false }, "2B.1.2": { "name": "Algorithm type", "category": "Algorithm – throughput", "type": "string", "description": "Why is this type of algorithm chosen?", "required": false }, "2B.1.3": { "name": "Algorithm type", "category": "Algorithm – throughput", "type": "string", "description": "Why is this type of algorithm best suited to achieve the objectives formulated 2B.1.4 What alternatives are there and why are they less appropriate or useful? in question 1.2?", "required": false }, "2B.1.4": { "name": "Algorithm type", "category": "Algorithm – throughput", "type": "string", "description": "What alternatives are there and why are they less appropriate or useful?", "required": false } , "2B.2.1": { "name": "Ownership and control", "category": "Algorithm – throughput", "type": "string", "description": "If the algorithm has been developed by an external party: have clear agreements been made on the ownership and management of the algorithm? What are those agreements?", "required": false } , "2B.3.1": { "name": "Algorithm accurac", "category": "Algorithm – throughput", "type": "string", "description": "What is the accuracy of the algorithm? On the basis of which evaluation criteria is this accuracy determined?", "required": false }, "2B.3.2": { "name": "Algorithm accurac", "category": "Algorithm – throughput", "type": "string", "description": "Is the level of accuracy (question 2B.3.1) acceptable for the way the algorithm will be will be used?", "required": false }, "2B.3.3": { "name": "Algorithm accurac", "category": "Algorithm – throughput", "type": "string", "description": "How is the algorithm tested?", "required": false }, "2B.3.4": { "name": "Algorithm accurac", "category": "Algorithm – throughput", "type": "string", "description": "What measures can be taken to counteract the risks of reproduction or even amplification of biases (e.g. different sampling strategy, feature modification, ...)?", "required": false }, "2B.3.5": { "name": "Algorithm accurac", "category": "Algorithm – throughput", "type": "string", "description": "What assumptions underlie the selection and weighting of the indicators? Are those assumptions valid? Why or why not?", "required": false }, "2B.3.6": { "name": "Algorithm accurac", "category": "Algorithm – throughput", "type": "string", "description": "How often is the algorithm wrong? (e.g. in terms of number of false positives, false negatives, R-squared, ...)", "required": false } , "2B.4.1": { "name": "Transparency and explainability", "category": "Algorithm – throughput", "type": "string", "description": "Is it clear what the algorithm does, how it does this, and on what basis (what data) it does this? Please explain.", "required": false }, "2B.4.2": { "name": "Transparency and explainability", "category": "Algorithm – throughput", "type": "string", "description": "For which people and groups (internal and external) will the operation of the algorithm be made transparent and how is this done?", "required": false }, "2B.4.3": { "name": "Transparency and explainability", "category": "Algorithm – throughput", "type": "string", "description": "For which target groups must the algorithm be explainable?", "required": false }, "2B.4.4": { "name": "Transparency and explainability", "category": "Algorithm – throughput", "type": "string", "description": "Can the operation of the algorithm be explained in a sufficiently understandable manner for the target groups identified in question B.4.3?", "required": false } , "3.1.1": { "name": "Decisions based on algorithmic output", "category": "Implementation, use and supervision – output", "type": "string", "description": "What happens with the results or outcome of the algorithm? What decisions are based on them?", "required": false } , "3.2.1": { "name": "The role of humans in the decision", "category": "Implementation, use and supervision – output", "type": "string", "description": "What role do humans play in decision-making on the basis of the algorithmic output? How are they empowered to make decisions responsibly on the basis of the algorithmic output?", "required": false }, "3.2.2": { "name": "The role of humans in the decision", "category": "Implementation, use and supervision – output", "type": "string", "description": "Is there sufficient qualified staff in place to manage, review and adjust the algorithm, if needed, and will there be in future?", "required": false } , "3.3.1": { "name": "Effects of the algorithm", "category": "Implementation, use and supervision – output", "type": "string", "description": "What will be the effects of using the algorithm for citizens and how will the ‘human measure’ be considered when making decisions based on the algorithm?", "required": false }, "3.3.2": { "name": "Effects of the algorithm", "category": "Implementation, use and supervision – output", "type": "string", "description": "What are the risks of stigmatisation, discrimination or otherwise harmful or adverse effects on citizens? How will these be combated or mitigated?", "required": false }, "3.3.3": { "name": "Effects of the algorithm", "category": "Implementation, use and supervision – output", "type": "string", "description": "How will the expected effects contribute to solving the problem that prompted the development/deployment of the algorithm (see question 1.1) and to achieving the proposed objectives (see question 1.2)?", "required": false }, "3.3.4": { "name": "Effects of the algorithm", "category": "Implementation, use and supervision – output", "type": "string", "description": "How do the expected effects relate to the values being served (see question 1.3)? How are risks of undermining certain values handled?", "required": false } , "3.4.1": { "name": "Procedures", "category": "Implementation, use and supervision – output", "type": "string", "description": "What procedures are in place for taking decisions on the basis of the algorithm?", "required": false }, "3.4.2": { "name": "Procedures", "category": "Implementation, use and supervision – output", "type": "string", "description": "How are the various relevant actors (administratively and politically accountable persons, citizens) engaged in decision-making?", "required": false }, "3.4.3": { "name": "Procedures", "category": "Implementation, use and supervision – output", "type": "string", "description": "How is it being safeguarded that these procedures meet the requirements of good governance, good administration and – where necessary – legal protection?", "required": false } , "3.5.1": { "name": "Context", "category": "Implementation, use and supervision – output", "type": "string", "description": "Time/period: when will the algorithm be used? For how long will it be used?", "required": false }, "3.5.2": { "name": "Context", "category": "Implementation, use and supervision – output", "type": "string", "description": "Area: where will the algorithm be used? Is it in a specific geographical area; does the algorithm concern a certain group of persons or cases?", "required": false }, "3.5.3": { "name": "Context", "category": "Implementation, use and supervision – output", "type": "string", "description": "Can the algorithm still be used if context factors change or if the algorithm is used in another context than the one for which it was developed?", "required": false } , "3.6.1": { "name": "Communication", "category": "Implementation, use and supervision – output", "type": "string", "description": "How open can you be about the operation of the algorithm in the light of the objectives and context of its deployment?", "required": false }, "3.6.2": { "name": "Communication", "category": "Implementation, use and supervision – output", "type": "string", "description": "How do you intend to communicate about the use of the algorithm?", "required": false }, "3.6.3": { "name": "Communication", "category": "Implementation, use and supervision – output", "type": "string", "description": "Will the algorithmic output be visualised, for example, in a table, graph, or dashboard? If so: is the shape of the visualisation or depiction a correct representation of the algorithmic output? Is the visualisation easy to read for various user groups?", "required": false } , "3.7.1": { "name": "Evaluation, auditing and safeguarding", "category": "Implementation, use and supervision – output", "type": "string", "description": "Have adequate tools been provided for evaluation, auditing and safeguarding of the algorithm?", "required": false }, "3.7.2": { "name": "Evaluation, auditing and safeguarding", "category": "Implementation, use and supervision – output", "type": "string", "description": "Are there sufficient opportunities to render proper account about the algorithm?", "required": false }, "3.7.3": { "name": "Evaluation, auditing and safeguarding", "category": "Implementation, use and supervision – output", "type": "string", "description": "What possibilities are there for auditors and regulators to attach (formal) consequences to the to the use of an algorithm by the government (e.g., feedback of findings, making recommendations, budgetary consequences, ...)", "required": false } , "4.1.1": { "name": "Fundamental right", "category": "Roadmap", "type": "string", "description": "Is any fundamental right affected by the algorithm that is to be used?", "required": false } , "4.2.1": { "name": "Specific legislation and standards", "category": "Roadmap", "type": "string", "description": "Do specific legal provisions or standards apply to the fundamental rights infringement?", "required": false } , "4.3.1": { "name": "Defining seriousness", "category": "Roadmap", "type": "string", "description": "How seriously is a fundamental right affected by the algorithm?", "required": false } , "4.4.1": { "name": "Objectives", "category": "Roadmap", "type": "string", "description": "Which objectives are pursued by using the algorithm?", "required": false } , "4.5.1": { "name": "Efficacy", "category": "Roadmap", "type": "string", "description": "Is the algorithm that is to be used a suitable means to realise the set objectives?", "required": false } , "4.6.1": { "name": "Necessity and subsidiarity", "category": "Roadmap", "type": "string", "description": "Is using this specific algorithm necessary to achieve this objective, and are there no other or mitigating measures available to do so?", "required": false } , "4.7.1": { "name": "Balancing interests/proportionality", "category": "Roadmap", "type": "string", "description": "Does the use of the algorithm result in a reasonable balance between the objectives pursued and the fundamental rights that will be infringed?", "required": false } } }