Parliament Resolution on Robotics, debated in 2017. Article 12 of this resolution says that it should always be possible to supply the rationale behind any decision taken with the aid of Artificial Intelligence (AI) that can have a substantive impact on one or more persons' lives. It must always be possible to reduce the AI system's computations to a form comprehensible by humans. Interestingly that same article articulates the necessity that 'advanced robots should be equipped with a 'black box' which records data on every transaction carried out by the machine, including (my italics CJ) the logic that contributed to its decisions'.8 This is in line with the regulation on the protection of natural persons with regard to the processing of personal data and on the free movement of such data which was adopted in April 2016 by the European Union. This regulation sets rules and requirements for data-driven automated processing. Every person has 'the right not to be subject to a decision which is based solely on automated processing (...)'.9 Examples that are explicitly mentioned are automatic refusal of an online credit application and e-recruiting practices without human intervention. Predictive profiling, one of the most fundamental intrusions in a private life which is increasingly used by authorities in fighting crime and terror remains allowed 'where expressly authorised by Union or Member State law'. Big data analysis is useful for revealing general patterns, but, and that is not always kept in mind, there is always a probability of a mismatch between general patterns and a specific situation (WRR, 2016, p. 27). Scholars, journalists, advisory and legislative bodies warn against excessive techno-dependency and techno-optimism. The Dutch investigative journalist Dimitri Tokmetzis criticises the naïve way rules are formulated and used in algorithms without paying enough attention to the validity of the underlying assumptions. To give an example: Dutch government assumes that poverty is a risk factor for the education of children. It is possible to design an algorithm to find evidence of poverty in the electronic child records and to make a list of families that need to be watched closely to be able to intervene if necessary. But do we know whether the assumption behind the rule is valid? Is the assumption that defines the rule based on thorough scientific research? (Tokmetzis, 2012, p. 59-60). Scott Mason, researcher at Keele University, also warns against the often-careless way how bureaucrats and policy-makers interpret and contextualise the results of Big Data itself without consulting domain experts to assess the validity of correlations. He is very critical about the claim that Big Data analysis creates the possibility for 'neutral' evidence based policy-making. According to Mason, 'the vast quantities of correlations generated by Big Data analytics act simply to broaden the range of 'evidence from which politicians can chose to support their arguments' (Mason, 2016). In 2016, the Dutch Scientific Council for Government Policy notified a highly undesirable tendency of policymakers to accept the revealed patterns without questioning the validity of the results for specific situations (WRR, 2016). Archival implications The aforementioned examples show that computational algorithms increasingly become an integrated part of government processes and decision making. Legal scholars have argued for more than 20 years in favor of more transparency in automated processing (Kroll, 2015, p. 6). Since records which are created in the course of business, 'provide evidence of actions, decisions, and intentions, both legal and illegal, proper and improper, and wise and misguided' (Cox Wallace, 2002, p. 4), availability of records is vital for accountability and transparency. Also in the recordkeeping realm, algorithmic tooling is used to manage the growing number of documents and to find relevant information for a specific purpose. The most advanced developments of algorithmic computation in the recordkeeping sphere can be found in eDiscovery and information retrieval applications.10 Since archivists claim to play a pivotal role in defending institutional and societal transparency and accountability (Jimerson, 2009, p. 246-252), there is an urgent need for archivists to ruminate what it means to take this role in the era of ubiquitous computing. If records, archives and archivists want to continue to be key players in ensuring and defending accountability, this evokes the question what meaningful recordkeeping is in this new context of data-ubiquity, and at the same time what meaningful records are. As Upward and others have put forward, this is exactly one of the main challenges the archival and recordkeeping community is confronted with: to clarify how the conceptual relationship between data, records and archives is designated in the era of ubiquitous computing. The traditional record was based on fixity and stability in a material sense. What has fundamentally changed is the possibility to produce different aggregates out of the same recorded data, which means, as Bruno Latour (2009) writes, 'that the whole has lost its privileged status' which makes us aware of the fact that the whole is always simpler than the parts (p. 198). The written record used to have the shape of an entity (the whole) in which the parts (words, sentences, paper, lay out, signature etc.) were a fixed materialised aggregate. Since the computational turn, it is possible to use the same recorded parts (data) in different configurations simultaneously. The stable whole has been replaced by a 'continually evolving liquid assemblage of action' (Introna, 2016, p. 19). It is as if we construct different types of houses with the same bricks at the same time. This (informational) fluidity is an important feature of what was designated by Deleuze as an assemblage. An assemblage is in the words of Deleuze 'a multiplicity which is made up of many heterogeneous terms and which establishes liaisons, relations between them [T]he assemblage's only unity is that of co-functioning' (Deleuze Parnet, 2007, p. 69). In an assemblage, an element can be dissociated from a specific assemblage and continue to function in another assemblage. Assemblages exist merely because of the relationships between the elements. Deleuze emphasises that an assemblage is never technological: '[t]ools always presuppose a machine, and the machine is always social before being technical. archives in liquid times 204 charles jeurgens threats of the data-flood. an accountability perspective in the era of ubiquitous computing. 8 European Parliament, Motion for a European Parliament resolution. Report with recommendations to the Commission on Civil Law Rules on Robotics, A8-0005/2017, art. 12, available at <http://www.europarl.europa.eu/sides/getDoc.do?pubRef=-//EP//TEXT+REPORT+A8-2017- 0005+0+D0C+XML+V0//EN> accessed at 30 March 2017. 9 Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation), art 71, available at <http://eur-lex. europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32016R0679&from=EN> accessed 30 March 2017. 10 The Information Governance Initiative Community provides an interesting overview of activities in these fields: http://iginitiative.com/community/ 205

Periodiekviewer Koninklijke Vereniging van Archivarissen

Jaarboeken Stichting Archiefpublicaties | 2017 | | pagina 104