TRAIF 2026—Communicating AI Ethics: In Research And Practice

Value of the Information & Strength of the Argument

VALUE OF INFORMATION: The presentations offer useful but not groundbreaking perspectives. Andrea Martin provides a clear, industry-oriented overview of AI governance, emphasizing practical implementation within a large corporation. The NIST framework example and the mainframe discussion add concrete elements. Boris Abramovic presents a thoughtful argument for using arts to improve public participation, supporting it with references to literature and examples from his lab. Loreta Tauginienė’s segment is more generic, likely describing procedural aspects without deep analysis. The arguments are logically structured, but the depth is limited due to the short format and multiple speakers. No conflicting viewpoints are presented, and all align with mainstream AI ethics discourse.

Scientific Rigor, Source Quality, Title Accuracy

SCIENTIFIC RIGOR AND SOURCE QUALITY: The speakers reference authoritative sources like NIST AI Risk Management Framework, OECD frameworks, EU AI Act, and GDPR, which are credible but not detailed. The description provides links to alignAI, IEAI, and the forum website, which are relevant institutional sources. However, no direct citations to specific studies are given in the transcript. The title matches the content. The session is part of a funded EU project (alignAI), which adds some credibility. Comment analysis: no comments are provided, so no public trends can be assessed.

The title accurately reflects the content: a forum session on communicating AI ethics in research and practice, covering governance frameworks, art-based public engagement, and ethical review processes.

7/10

The talks are delivered by experienced professionals and academics (IBM, University of Vienna, Kaunas University of Technology). They reference established frameworks (NIST, OECD, EU AI Act) and academic literature, but no empirical data or detailed citations are provided in the transcript. The technical depth varies, with some parts are more conceptual than hands-on.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 02:00Andrea Martin begins presenting on AI ethics as an enterprise value at IBM, discussing the shift from trustworthiness to AI governance.
• 09:00She explains the three IBM principles (augmented intelligence, data ownership, transparency) and five pillars of responsible AI.
• 17:00Discussion on the IBM mainframe as an example of embedded AI governance features, including synthetic data and drift detection.
• 23:00Boris Abramovic starts his talk on art-based approaches for public participation in AI ethics, referencing critiques of conventional participation.
• 35:00He presents examples from the Experimental Game Cultures lab, like facial recognition games and ChatGPT role-play, emphasizing emotional and future-oriented engagement.
• 44:00Loreta Tauginienė begins her presentation on ethical oversight by research ethics committees and the communication of approval processes.

• alignAI — EU Horizon project funding and information mentioned in description.
• Institute for Ethics in Artificial Intelligence (IEAI) — Organizer of the forum, referenced in description.
• IEAI Newsletter — Subscription link for updates, provided in description.
• Responsible AI Forum — Official website of the forum, linked in description.
• Amerikahaus — Venue provider, credited in description.
• IEAI Events — Page for other IEAI events, linked in description.

• NIST AI Risk Management Framework — Used by IBM and consistent with best practices in AI governance.
• EU AI Act — Mentioned as a compliance driver, aligning with the sector's regulatory direction.

• Anthropic's claim on mainframe COBOL translation — Andrea Martin disputes the idea that mainframes are obsolete and easy to replace, suggesting that such claims overlook business integration complexities.