Rutherford House, RHMZ03

*Note different room!*

A summary of what’s in the 2025 International AI Safety Report

In this seminar I'll summarise the main themes in the International AI Safety Report, which was commissioned at the Bletchley Park summit, and presented at the Paris summit. This report was basically ignored in the leaders' communiqué, which many see as the summit's main failing.. but we won't ignore it in our seminars!

New Tech, Old Tactics: Facial Recognition and the Policing of Māori

Facial recognition technology (FRT) is increasingly utilised within New Zealand's justice realm despite international trends where its application disproportionately impacts Indigenous and racialised communities. While being marketed as ‘objective’ and ‘race-neutral’, FRT often struggles with accuracy for darker skin tones and Indigenous features like facial moko/tattoos. These challenges raise important questions about Māori Data Sovereignty, tikanga, tapu, and the ethical implications of using Māori faces in global databases without consent, particularly amidst limited regulation and consultation with Māori communities.

This kōrero draws on my current PhD research, exploring how FRT intersects with historical practices of surveillance and control under settler colonialism. By analysing its use in policing, government services, and retail settings, I highlight the importance of transparency, accountability, and community engagement in the deployment of such technologies - especially for Māori here in Aotearoa. This discussion invites reflection on how we can ensure FRT aligns with equitable societal values and consider whether some technologies might perpetuate inequities rather than resolve them.

In this seminar Chris will explore the topic of AI harm through the use of critical theory, legal theory and analogy. The aim of the seminar is to utilise a multi-disciplinary approach to contextualise AI harm, and to illustrate that though a number are in fact new and novel, many are topics which have been grappled with in the humanities and legal fields for years.

Laby Building, LBLT118

Note this talk is at the Kelburn campus!

Whither artificial intelligence? Science, government, industry and society

Koi Tū, and its predecessors have long been engaged in considering the impacts of digital technologies on society. In our work for the OECD in the going digital project 2015-16, we took the lead on exploring the impact on institutions, of self, social and civil life. We tried to apply that work in the issues that surrounded the initial steps government took towards big data with the IDI, but the essential need for an independent oversight mechanism, which was absent remains, and the need has escalated. Through my roles at the International Science Council, Hema Sridhar and I developed a framework to assist policy makers in dealing with rapidly emerging technologies such as AI – a framework which has been part of discussions from Geneva to New York.
It was during the Science System Advisory Group’s deliberations that we have come to focus on several questions which will be the primary focus on this presentation. We start with the observation that in advanced technologies we will never be fully sovereign, and strategic partnerships will be necessary both within and beyond our shores. First science itself and education are being changed massively by AI and we are only starting on the journey. Secondly the government itself will be a large and growing user of AI. The use of AI on policy development is rapidly appearing overseas, but the systems, oversight and training needs a whole of government approach. There is much that can be achieved with big data in policy making and in evidence review that could make a large difference to the efficiency and effectiveness of government. It goes well beyond simply services provision. Thirdly, in the defense and security space including the financial sector the potential of AI and quantum to change the whole basis of these sectors is real and we are poorly prepared relative to other countries. Fourthly, the potential of advanced technologies such as AI to enhance current industries and to foster new segments is high. But our national investment has been poor and not and strategically coordinated. In sum much of our activity is a set of individual activities and the need to find ways to bring capacities and skills together without stifling competition and innovation is at the heart of the Science System Advisory Group’s two reports and the recommendations that we will make as the process concludes. Exploiting AI cannot be seen differently, at least in principle from any other technological research and exploitation path. The role of social sciences in ensuring its appropriate use is key.

AI Across the Pond 1: AI and Communications: New Machines and New Concepts

In this presentation I will consider ways in which the rapidly growing use of Artificial Intelligence (AI) by consumers as well as industry is transforming how we think about communications as both a social practice and as an academic field. I will give particular attention to three issues: the changing status of machines as communications actors and not simply platforms for human communication; the implications for trust from a growing reliance upon automated information and decision-making systems; and how global communications as a field of study may be challenged by the changing relationship between data and geopolitics.

New Zealand's AI Strategy: A first look

The NZ government has just released its Strategy for AI. In this seminar, we will give an overview of the strategy document, and the context of its creation. This will be an early opportunity to talk about the strategy, and compare to strategies developed in other countries.

• A few links from Tom Barraclough: Brainbox's New Zealand AI Policy Tracker; a piece about Sovereign AI for New Zealand; some thoughts on NZ's AI Strategy. (All presented in the DocRef format.)
• A link from Simon McCallum, to his interview on The Platform

Dispelling the Illusion of Learning: Shifting the Culture for Professional Students in an AI Future

As generative AI systems increasingly match (and at times outperform) the best student work in common university assessment tasks such as analytical writing, problem-solving, and applied reasoning, educators face a fundamental question: what kinds of learning still matter, how should we teach for them, and how should we assess them?

This talk brings together brings together perspectives from the Faculty of Law and the Wellington School of Business and Government to explore how AI is transforming professional education. Drawing on recent experiments and first-hand teaching experience, we offer tentative analyses of the growing capability of AI to perform traditional academic and professional tasks and what this means for assessment, curriculum design - and more broadly the future of knowledge work and 'the university'.

We will consider to what degree students still need to demonstrate these capabilities to become experts who can use AI as a lever for higher-order reasoning, or whether AI's growing competence demands a sharper pivot: toward helping students identify, apply, and extend the kinds of judgment, contextual sensitivity, and ethical reasoning that remain distinctly human.

We will also consider the danger of the persistent illusion of learning - the assumption that task completion signals true understanding - and explore how AI reveals its limits.

The impact of AI on assessments and student learning at University

The Royal Society Te Apārangi's guidelines on Generative AI in Research: A first report, and some ongoing work

In this session, we will introduce some guidelines recently published by the Royal Society Te Apārangi, on Uses of Generative AI for research in Aotearoa New Zealand. We will also have a discussion about how more detailed guidelines could be provided, that cater for different research disciplines and different research tasks, while keeping up with the rapidly changing state of AI technology.