The Limits of Value Transparency in Machine Learning
| cam.depositDate | 2022-05-04 | |
| cam.issuedOnline | 2022-06-13 | |
| cam.orpheus.counter | 24 | |
| cam.orpheus.success | Mon Jan 23 08:37:36 GMT 2023 - This item is covered by RRS with an embargo. The item is now published and embargo has been lifted. | |
| dc.contributor.author | Nyrup, Rune | |
| dc.contributor.orcid | Nyrup, Rune [0000-0002-9880-6912] | |
| dc.date.accessioned | 2022-05-04T23:30:37Z | |
| dc.date.available | 2022-05-04T23:30:37Z | |
| dc.date.issued | 2022 | |
| dc.date.updated | 2022-05-04T11:07:41Z | |
| dc.description.abstract | <jats:title>Abstract</jats:title><jats:p>Transparency has been proposed as a way of handling value-ladenness in machine learning (ML). This article highlights limits to this strategy. I distinguish three kinds of transparency: epistemic transparency, retrospective value transparency, and prospective value transparency. This corresponds to different approaches to transparency in ML, including so-called explainable artificial intelligence and governance based on disclosing information about the design process. I discuss three sources of value-ladenness in ML—problem formulation, inductive risk, and specification gaming—and argue that retrospective value transparency is only well-suited for dealing with the first, while the third raises serious challenges even for prospective value transparency.</jats:p> | |
| dc.description.sponsorship | This research was funded in whole, or in part, by the Wellcome Trust [Grant number 213660/Z/18/Z] and the Leverhulme Trust, through the Leverhulme Centre for the Future of Intelligence. | |
| dc.identifier.doi | 10.17863/CAM.84161 | |
| dc.identifier.eissn | 1539-767X | |
| dc.identifier.issn | 0031-8248 | |
| dc.identifier.uri | https://www.repository.cam.ac.uk/handle/1810/336740 | |
| dc.language.iso | eng | |
| dc.publisher | Cambridge University Press (CUP) | |
| dc.publisher.department | Leverhulme Centre For The Future of Intelligence | |
| dc.publisher.url | http://dx.doi.org/10.1017/psa.2022.61 | |
| dc.rights | Attribution 4.0 International | |
| dc.rights.uri | https://creativecommons.org/licenses/by/4.0/ | |
| dc.subject | 5003 Philosophy | |
| dc.subject | 50 Philosophy and Religious Studies | |
| dc.subject | 5002 History and Philosophy Of Specific Fields | |
| dc.subject | Clinical Research | |
| dc.title | The Limits of Value Transparency in Machine Learning | |
| dc.type | Article | |
| dcterms.dateAccepted | 2022-04-14 | |
| prism.publicationName | PHILOSOPHY OF SCIENCE | |
| pubs.funder-project-id | Wellcome Trust (213660/Z/18/Z) | |
| pubs.funder-project-id | Leverhulme Trust (RC-2015-067) | |
| pubs.funder-project-id | Leverhulme Trust (RC-2015-067) | |
| pubs.licence-display-name | Apollo Repository Deposit Licence Agreement | |
| pubs.licence-identifier | apollo-deposit-licence-2-1 | |
| rioxxterms.type | Journal Article/Review | |
| rioxxterms.version | AM | |
| rioxxterms.versionofrecord | 10.1017/psa.2022.61 |
Files
Original bundle
1 - 1 of 1
Loading...
- Name:
- LimValTransp_Final+cover_page.pdf
- Size:
- 161.8 KB
- Format:
- Adobe Portable Document Format
- Description:
- Accepted version
- Licence
- https://creativecommons.org/licenses/by/4.0/