real-time sharing of information isn't possible yet
supervised learning is also not done yet
information position of citizens could be improved by sharing more information pro-actively
citizens are not necessarily aware that an adverse decision was made in part by an algorithm
should align with existing mechanisms -- complaints procedure and if that does not work ombudsman
reporting point for chain errors (``meldpunt voor keten fouten'')
    incident driven
other mechanism is object to decisions
    this is the formal legal procedure ``bezwaar/beroep''
    objection committee needs to have knowledge and information to evaluate role of algorithms in objections
    case file should include all relevant information about algorithm that was used:
        ``But what you see then what we actually ask for documentation and explanation about procedural and technical transparency to explain the model, the technical operation, the data sets any bias everything that has to land there so that they are not things that used to be not necessarily things that appear in your case file all relevant facts and circumstances that led to a decision, but if it means if you have used a signaling algorithm for, for example, a prior check, you will actually if that is a question in that assessment of the notice of objection of how are you doing so all those extra documents it is very official what I am saying now but you should produce it and there should be a judgment about it and that committee then advises the administrative body to take a decision of yes that was right or no that was not right and we would have had a different outcome yes or no without the algorithm.''
    ``beroep'' (appeal) is what comes after objection, this is handled by a judge
everything depends on information position of citizens
informal arenas could help educate citizens
used citizen panels to determine what should go into algorithm register
citizen participation is hard because it is a struggle to find people and companies who are willing and able to contribute
    counterargument is also that government should just do their job well and not depend on individual citizens so much
also make use of external commission CPA -- when personal data is processed or a system is high risk, getting their advice is mandatory
models are always a simplification of reality
individual complaints should lead to system changes if necessary
    in addition to individual redress or compensation
participation during development
    human rights impact assessments could be opened up
    bias analyses and fundamental rights considerations as well
    opened up for debate
still, would be hard to get citizens to participate on equal footing and representatively
    this is why academia and civic experts are important
decisions to develop AI or take into use AI should also be political, debated in city council
    this takes care of representation