Until now, it has been the case that the use of entirely new technologies has been at least publicly debated and democratically legitimized and democratically legitimized by a law. This is not the case with the AMS algorithm.
This system was developed and presented without public debate. Civil society was not involved in this process, and to this day there is a fundamental lack of transparency about how this system works and decides about us.
A similar system in Poland was overturned by the Constitutional Court overturned. In the Netherlands, too, an algorithm-based system, that was supposed to monitor welfare recipients was declared unlawful and unlawful and was discontinued.
We must not allow characteristics, characteristics over which we have no control be decisive in determining whether people are Education and training are prohibited. Algorithmic Systems must combat prejudice and injustice and not and must not cement these injustices cement these injustices in place.
People should be able to make decisions about to decide on access to government services, instead of just nodding off decisions made by a computer.
The AMS algorithm must be rejected and shut down. A computer must not be allowed to decide human fates, and access to state benefits must not be regulated by a computer. Decisions must be made comprehensibly, transparently, and with human measure. The system is a dangerous precedent for outsourcing political decisions to algorithms.
Even if the system does not (yet) make any fully automated decisions, it must not prescribe any decisions that the AMS staff only nods off. The AMS counseling sessions usually last only 15 minutes, and there is enormous pressure to justify not contradicting the computer.
Enabling more education, by promoting and enabling (further) education
People should be promoted on the basis of their skills and strengths, and this requires an increase in further education and training measures for job seekers.
Jobseekers must be allowed to start university or UAS studies (this is currently only allowed if jobseekers have already studied in parallel during their working lives). The AMS should also allow self-paid continuing education, e.g. courses lasting only a few days.
We demand more and better trained counselors who have enough time for counseling.
The ratio of counselors to job seekers in Germany is 1:100, while in Austria it is only 1:250. This shows that an increase in this area is conceivable and would make sense, because a better counseling ratio means better job opportunities and higher-quality counseling. The AMS not only needs more money for the training and further education of its consultants. It is equally necessary to increase the quality of meaningful further education and training measures for job seekers.
We call for comprehensive proactive education about job seekers' rights.
The AMS is funded by all of us and is therefore a service provider with a mandate to help the general public. It must meet all people at eye level. In order for people to be able to talk to their advisors on an equal footing, they first of all need information about what their rights and obligations are in connection with the AMS; starting with what their entitlement to unemployment benefits is linked to - whether and under what circumstances their money can be cut off - whether they have to enter into a support agreement with the AMS or whether this is optional - what their rights to information and disclosure look like and how the legal relationship between them and external partner companies of the AMS is regulated.
Projects and programs that are started with all of our money must be as transparent as possible in order to be evaluable.
The AMS algorithm has been developed for years at great expense, with the public excluded. Informed debate has been impossible simply because neither the system's database nor its results have been made public. Training materials for consultants and processes for appealing the algorithm continue to be kept secret. A scientific reappraisal and democratic debate about the project are thus excluded.
We demand comprehensive and proactive transparency from the AMS. Clarification must not come only through public pressure. At the very least, the mathematical models, the software used, the anonymized data basis, the concrete functioning of the system and the results of the calculation process for individual groups must be published. Likewise, it must be made transparent how the AMS algorithm affects practical counseling and the allocation of subsidies. The measures that the AMS has already taken as a result of the criticism, such as training materials, the interaction with the restructuring of the AMS and accompanying processes, as well as complaint options for those affected, must also finally be published.
We call for a review of algorithms for their technical and social impact before they are introduced (obligation for an upstream Algorithmic Impact Assessment).
Automated systems that could have negative consequences for people must be closely scrutinized for their functionality and possible consequences before they are introduced. This must be done for all algorithms that potentially interfere with fundamental rights
Here there are international best practices for the control of algorithms, a so-called Algorithmic Impact Assessment. This was neglected in the case of the AMS algorithm and must be mandatory in the future.
Algorithms have to be checked for errors in the system, biases in the data and resulting and resulting negative feedback cycles must be analyzed, before such systems are launched and widely deployed
Not every statistical regression follows the principle of cause and effect Effect. Negative effects on people due to incorrect analyses must be be avoided
That is why we are calling for an Algorithmic Impact Assessment, as is already the case in Canada or the City of New York have already introduced.
We demand optional participation in the AMS algorithm (opt-in) and decoupling from any grant award. We want a system that is so good that the people concerned participate in it voluntarily
The AMS is in a position of power vis-à-vis job seekers In the interest of data protection, we therefore call for the voluntary participation of all jobseekers in the AMS algorithm of all job seekers in the AMS algorithm. Job seekers must give their informed consent before the system categorizes them. This Consent can be revoked at any time, which also requires that all calculated Values of the algorithm must be deleted. From the refusal to participate in this System, the jobseekers must not suffer any disadvantages.
People should be able to make decisions about to decide on access to government services, instead of just nodding off decisions made by a computer.
The AMS algorithm must be rejected and shut down. A computer must not be allowed to decide human fates, and access to state benefits must not be regulated by a computer. Decisions must be made comprehensibly, transparently, and with human measure. The system is a dangerous precedent for outsourcing political decisions to algorithms.
Even if the system does not (yet) make any fully automated decisions, it must not prescribe any decisions that the AMS staff only nods off. The AMS counseling sessions usually last only 15 minutes, and there is enormous pressure to justify not contradicting the computer.
Enabling more education, by promoting and enabling (further) education
People should be promoted on the basis of their skills and strengths, and this requires an increase in further education and training measures for job seekers.
Jobseekers must be allowed to start university or UAS studies (this is currently only allowed if jobseekers have already studied in parallel during their working lives). The AMS should also allow self-paid continuing education, e.g. courses lasting only a few days.
We demand more and better trained counselors who have enough time for counseling.
The ratio of counselors to job seekers in Germany is 1:100, while in Austria it is only 1:250. This shows that an increase in this area is conceivable and would make sense, because a better counseling ratio means better job opportunities and higher-quality counseling. The AMS not only needs more money for the training and further education of its consultants. It is equally necessary to increase the quality of meaningful further education and training measures for job seekers.
We call for comprehensive proactive education about job seekers' rights.
The AMS is funded by all of us and is therefore a service provider with a mandate to help the general public. It must meet all people at eye level. In order for people to be able to talk to their advisors on an equal footing, they first of all need information about what their rights and obligations are in connection with the AMS; starting with what their entitlement to unemployment benefits is linked to - whether and under what circumstances their money can be cut off - whether they have to enter into a support agreement with the AMS or whether this is optional - what their rights to information and disclosure look like and how the legal relationship between them and external partner companies of the AMS is regulated.
Projects and programs that are started with all of our money must be as transparent as possible in order to be evaluable.
The AMS algorithm has been developed for years at great expense, with the public excluded. Informed debate has been impossible simply because neither the system's database nor its results have been made public. Training materials for consultants and processes for appealing the algorithm continue to be kept secret. A scientific reappraisal and democratic debate about the project are thus excluded.
We demand comprehensive and proactive transparency from the AMS. Clarification must not come only through public pressure. At the very least, the mathematical models, the software used, the anonymized data basis, the concrete functioning of the system and the results of the calculation process for individual groups must be published. Likewise, it must be made transparent how the AMS algorithm affects practical counseling and the allocation of subsidies. The measures that the AMS has already taken as a result of the criticism, such as training materials, the interaction with the restructuring of the AMS and accompanying processes, as well as complaint options for those affected, must also finally be published.
We call for a review of algorithms for their technical and social impact before they are introduced (obligation for an upstream Algorithmic Impact Assessment).
Automated systems that could have negative consequences for people must be closely scrutinized for their functionality and possible consequences before they are introduced. This must be done for all algorithms that potentially interfere with fundamental rights
Here there are international best practices for the control of algorithms, a so-called Algorithmic Impact Assessment. This was neglected in the case of the AMS algorithm and must be mandatory in the future.
Algorithms have to be checked for errors in the system, biases in the data and resulting and resulting negative feedback cycles must be analyzed, before such systems are launched and widely deployed
Not every statistical regression follows the principle of cause and effect Effect. Negative effects on people due to incorrect analyses must be be avoided
That is why we are calling for an Algorithmic Impact Assessment, as is already the case in Canada or the City of New York have already introduced.
We demand optional participation in the AMS algorithm (opt-in) and decoupling from any grant award. We want a system that is so good that the people concerned participate in it voluntarily
The AMS is in a position of power vis-à-vis job seekers In the interest of data protection, we therefore call for the voluntary participation of all jobseekers in the AMS algorithm of all job seekers in the AMS algorithm. Job seekers must give their informed consent before the system categorizes them. This Consent can be revoked at any time, which also requires that all calculated Values of the algorithm must be deleted. From the refusal to participate in this System, the jobseekers must not suffer any disadvantages.
Bundesministerium für Arbeit, Familie und Jugend
Untere Donaustraße 13-15, 1020 Wien
Once a year you can request all your data from all companies and authorities for free. Here you can do that for the AMS.
Request for information according to Art. 15 DSGVO
Only if you have been unemployed since 2019, the AMS has an algorithmic classification of you, which is about 2 pages of the information request (an info page and an information page). On average, a request for information at the AMS (for older persons) has about 80-150 pages. Upon request, your AMS counselorcan also provideyou with information about your algorithmic classification.