Diana Acosta

This paper examines when it is impermissible to outsource bureaucratic decisions to machine learning systems. It distinguishes between institutional legitimacy and decision permissibility. The former refers to the authority of an institution whose decisions have impact on matters of public interest. The latter refers to the internal features of automated decision-making procedures that make it permissible to substitute organizational decision-making with automated processes. The paper argues that the violation of certain criteria makes it impermissible to outsource a decision to a machine learning system. These criteria include five necessary, but not sufficient, conditions: due appreciation, consistency, justifiability, accuracy, and contestability. Failing any of these conditions renders a machine learning decision impermissible. Lastly, it explores the possibility that meeting all conditions might still not suffice in more complex scenarios.

The digital public forum has challenged many of our normative intuitions and assumptions. Many scholars have argued against the idea of free speech as a suitable guide for digital platforms’ content policies. This paper has two goals. Firstly, it suggests that there is a version of the free speech principle which is suitable for platforms that have adopted a commitment to free speech to guide their content curation strategies. I call it the Principle of Epistemic Resilience. Secondly, it aims to analyze some of the practical implications of the principle. It argues that upholding this principle in the digital public forum requires a comprehensive strategy, including (1) the automated removal and demotion of contents that threaten to cause serious harm; (2) changes to engagement optimization algorithms; and (3) changes to affordances inside the platform. These changes are necessary to create a fertile environment for deliberation, which is crucial to epistemic resilience. If such a comprehensive strategy is absent, platforms may actively undermine the societal value of speech.

Using a design-based research approach, we studied ways to advance opportunities for children and families to engage in engineering design practices in an informal educational setting. 213 families with 5–11-year-old children were observed as they visited a tinkering exhibit at a children’s museum during one of three iterations of a program posing an engineering design challenge. Children’s narrative reflections about their experience were recorded immediately after tinkering. Across iterations of the program, changes to the exhibit design and facilitation provided by museum staff corresponded to increased families’ engagement in key engineering practices. In the latter two cycles of the program, families engaged in the most testing, and in turn, redesigning. Further, in the latter cycles, the more children engaged in testing and retesting during tinkering, the more their narratives contained engineering-related content. The results advance understanding and the evidence base for educational practices that can promote engineering learning opportunities for children.

One of the central limitations of sensory substitution devices (SSDs) is their inability to reproduce the non-sensory feelings that are normally associated with visual experiences, especially hedonic and aesthetic responses. This limitation is sometimes reported to cause SSD users frustration. To make matters worse, it is unclear that improvements in acuity, bandwidth, or training will resolve the issue. Yet, if SSDs are to actually reproduce visual experience in its fullness, it seems that the reproduction of non-sensory feelings will be of some importance. We offer a novel solution. Researchers can produce hedonic and aesthetic responses by eliciting these feelings artificially, pairing distal objects that should be pleasurable to pleasurable stimulus outputs from the SSD. We outline two strategies for accomplishing this: first, by means of a prefixed, hardwired, association of pleasant distal objects to pleasant stimulus outputs from the SSD; second, by means of a flexible, feedback-based association which creates associations based on a subject-directed matching of distal objects to patterns of stimuli from the SSD which the subject takes to have the corresponding hedonic properties. We evaluate some problems with both strategies, and we argue that the feedback-based strategy is more promising. Researchers can use this strategy to help the blind, allowing them to take pleasure in the objects they perceive using SSDs.