Algorithmic Decision-Making Based on Machine Learning from Big Data: Can Transparency Restore Accountability?
Open accessAbstract
Decision-making assisted by algorithms developed by machine learning is increasingly determining our lives. Unfortunately, full opacity about the process is the norm. Would transparency contribute to restoring accountability for such systems as is often maintained? Several objections to full transparency are examined: the loss of privacy when datasets become public, the perverse effects of disclosure of the very algorithms themselves ("gaming the system" in particular), the potential loss of companies' competitive edge, and the limited gains in answerability to be expected since sophisticated algorithms usually are inherently opaque. It is concluded that, at least presently, full transparency for oversight bodies alone is the only feasible option; extending it to the public at large is normally not advisable. Moreover, it is argued that algorithmic decisions preferably should become more understandable; to that effect, the models of machine learning to be employed should either be interpreted ex post or be interpretable by design ex ante.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Transparency and accountability may be at odds in the case of algorithms.
“de Laat (2018) argues that transparency and accountability may even be at odds in the case of algorithms.”
From Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward (2020) · cites this paper as de Laat (2018)
Full transparency has four main objections: leaking of privacy sensitive data into the open; backfiring into an implicit invitation to game the system; harming company property rights with negative consequences on competitiveness; and inherent opacity of algorithms.
“Hence, he argues against full transparency along four main lines of reasoning: (i) leaking of privacy sensitive data into the open; (ii) backfiring into an implicit invitation to game the system; (iii) harming of the company property rights with negative consequences on their competitiveness (and on the developers reputation as discussed above); (iv) inherent opacity of algorithms, whose interpret”
From Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward (2020) · cites this paper as de Laat (2018)
In this setting, seeking algorithm transparency would not be harmful for their developers as scrutiny would be delegated to entrusted intermediate parties, to take place behind closed doors.
“In this setting, seeking algorithm transparency would not be harmful for their developers as scrutiny would be delegated to entrusted intermediate parties, to take place behind closed doors (de Laat, 2018).”
From Ethical principles in machine learning and artificial intelligence: cases from the field and possible ways forward (2020) · cites this paper as de Laat, 2018
Simpler code would arguably make it easier to game the system and would provide industry with fewer incentives to innovate.
“This would arguably make it easier to game the system, and it would provide the industry with fewer incentives to innovate (e.g., Zarsky 2016; Lepri et al. 2017;De Laat 2018; Carabantes 2019).”
From Artificial intelligence, transparency, and public decision-making: Why explanations are key when trying to produce perceived legitimacy (2020) · cites this paper as De Laat 2018
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 33 · Cited by 251
Research works printed but not linked: 2
- Kroll JA Huey J Barocas S Accountable algorithms University of Pennsylvania Law Review 2017 165 3 633 705
- Zarsky TZ Transparent predictions University of Illinois Law Review 2013 2013 4 1503 1570
- Authors
- de Laat, P. B.
- ORCID
- de Laat, Paul B.
- Venue
- Philosophy & Technology
- Published
- 2017-11-12
- Volume
- 31
- Issue
- 4
- Pages
- 525-541
- ISSN
- 2210-5433 (ISSN-L)
- 2210-5441
- PMC
- PMC6390893
- OpenAlex
- W2767837247
- Cited by
- 251
- References
- 33
- Citations by year
- 2026: 14 · 2025: 35 · 2024: 32 · 2023: 25 · 2022: 46 · 2021: 49 · 2020: 32 · 2019: 13 · 2018: 4
- FWCI
- 30.76
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Philosophy & Technology (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, other-oa)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Funders
- University of Groningen
- Topics
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Social Sciences › Law › Digitalization, Law, and Regulation
- Keywords
- algorithmic decision-making, machine learning, transparency, accountability, algorithmic opacity, privacy, gaming the system, competitive advantage, algorithmic accountability, explainable AI, interpretability, big data, algorithmic oversight
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
