At NGI Forum 2023, the panel Ethics in (NGI) search addressed not only the problems of contemporary internet search, namely Google, but forthcoming Open Source alternatives. The panel’s moderator, Mirko Presser (Aarhus University), first introduced a couple of projects funded by NGI search.
HeReFaNMi (Health-Related Fake News Mitigation) combats the propagation of fake news within health-related information and attempts to restore trustworthiness to the internet community.
SVP64 takes existing search algorithms and optimises a Vector ISA (at the hardware level) to increase energy efficiency.
Tech philosopher Fabian Geier (Code University) contributed a short overview of some of the ethical considerations regarding the awarded projects and emphasised the long-term considerations along with how ethics should be a positive and innovative force, rather than mere restriction.
Renée Ridgway (Aarhus University) provided examples of how Google’s business model has increased the flow of disinformation in an era of surveillance capitalism, whilst debunking its ‘personalisation’ to show how Google organises us(ers) into groups or categories of similar users when delivering its search results.
Victor Miloshevski (University of Balearic Islands) discussed issues of biased AI, putting forth a modernised ‘social contract’ à la Rousseau as a solution. Yet as noted during the ensuing discussion, Acton’s ‘power tends to corrupt, and absolute power corrupts absolutely’ when applied to the monopoly problem, with its ability to scale, doesn’t just disappear. However, the ongoing anti-trust lawsuit against Google in the US in not without merit and it can be an opportunity to bring about change––by breaking up the monopoly.
Alexander Nussbaumer (Graz University of Technology) introduced the EU web search alternative (OpenWebSearch.eu), an open web index that will enable an ecosystem of search applications for specific purposes and data products. Funded from September 2022 - August 2025 by the European Commission under the Horizon programme, this index will be facilitated and maintained by a consortium of 14 research partners and computer centres from seven European countries. Alongside of the technical and legal developments, the Working Group Ethics, including Nussbaumer and Ridgway and organised by Christine Plote (Open Search Foundation), meets every month online as well as conducting in-person workshops that explore a range of ethical issues regarding the building of such an index. For this panel, Alexander explained the work-in-progress ‘Ethics (STS) Methodology’ diagram, which refers to the discipline of Science and Technology Studies and offers a way to help visualise, structure and think-through the ethical processes and dilemmas concerning the forthcoming Open Web Index.
This working diagram is divided into two phases, the ‘Index Generation’ and the ‘Index Usage’. This is subdivided by two horizontal sections, the ‘Analysis’ stage and the ‘Solution’ stage. Within the ‘Analysis’ stage, the Working Group Ethics raises ethical issues through the format of a ‘Structured Analysis’ and a 'General Analysis'. The former is an ethical reflection for the step-by-step technological development of the index. With the latter, a general ethical discussion on search engines is undertaken. Simultaneously, the issues are translated into ethical values that are then taken up within the consortium of research partners, creating an ‘ethics-by-design’ approach for including feedback from the Working Group Ethics into the index generation. Additionally, a ‘Code of Ethics’ or ‘best practices’ will be made public for those wishing to use the index for the development of search applications and training large language models (LLMs). Moreover, workshops and discussions will be carried out to initiate discourse and awareness about ethical search at all stages along the way.
During the panel, questions were raised by the audience concerning how to deal with the rise of mis/dis/information and whether there could be ‘truth anchors’ that could somehow be incorporated into future search technologies. As explained by members of the Open Search Foundation present, the forthcoming Open Web Index will have labelling and user-friendly explanations containing the provenance, or source of the search results that are analogous to ‘nutrition labels’ for food products. Along with other filtering mechanisms, these types of innovations convey to the public some of the Open Web Index’s core values: openness, transparency, trustworthiness, privacy, responsibility and accountability, sustainability and reliability, to name only a few.