As New Zealand prepares for its first general election in the age of generative AI, independent media coverage is drawing attention to a critical question: how well are political parties represented when voters turn to AI for answers?
On 7 October 2026, Stuff journalist Finn Hogan published an in-depth analysis titled The first AI election is here, so far, everyone is losing.
The article featured research from the CiteCompass NZ Election 2026 AI Visibility Monitor, our independent 12-week longitudinal study examining how six parliamentary political parties and their policies appear in AI-generated answers.
The coverage moves an important discussion into the national spotlight: as AI increasingly mediates how people discover and evaluate information, are voters receiving a sufficiently complete picture of the policies available to them?
For CiteCompass, it is encouraging to see this emerging issue receiving independent journalistic scrutiny.
The First AI Election: A New Information Battleground
Political communication has evolved through successive technological shifts — television, traditional search engines, social media and now generative AI.
Each transition has changed how information reaches the public.
Unlike traditional search, where users generally receive a list of websites to explore, AI platforms increasingly deliver synthesised answers assembled from multiple sources.
A voter asking about housing affordability, healthcare, economic growth or public safety may receive a detailed explanation without ever visiting a political party’s website.
That creates a new consideration for political information: being published online does not automatically mean being represented in an AI-generated answer.
Stuff’s analysis explores this transition and the implications for political parties, voters and democratic debate.
What the CiteCompass Research Is Revealing
Our study monitors six parliamentary political parties across five policy domains:
- Cost of Living
- Economy, Jobs & Growth
- Health
- Housing
- Crime & Public Safety
Together, these represent 30 party-policy profiles monitored consistently over 12 weeks.
The research examines whether official policy information appears in AI-generated answers, how prominently it is represented, which sources are being cited, and how visibility changes over time.
When Stuff examined the Week 6 findings, an important pattern had emerged: measured AI visibility remained low across many of the monitored policy profiles.
Housing was particularly notable, with all six monitored profiles recording visibility scores below 10 out of 100 in Week 6.
These findings raise a question about the ability of AI search systems to discover and represent primary political information during voters’ research journeys.
The scores do not measure the quality, feasibility or desirability of any policy. They measure its visibility within the specific AI research framework.
Explore the CiteCompass research methodology
The Information Architecture Problem
One of the most interesting aspects of Stuff’s coverage was its examination of what prevents AI systems from discovering and interpreting information effectively.
Traditional Search Engine Optimisation (SEO) has shaped digital publishing for decades. Generative Engine Optimisation (GEO), also known as Answer Engine Optimisation (AEO), introduces additional considerations.
Information that is easy for a human to read may not be equally accessible to AI-powered search systems.
For example, important evidence may be embedded in images, policy information may be spread across multiple disconnected pages, or an established policy URL may change without an appropriate redirect.
AI systems also need sufficient context to interpret the significance of the information they discover.
These challenges are not unique to political parties. They are increasingly relevant to businesses, public institutions and any organisation seeking to communicate authoritative information online.
The underlying principle is straightforward: information needs to be accessible, clearly structured and supported by credible sources if it is to be reliably discovered and represented.
Can AI Improve Political Understanding?
The Stuff article also explores a more fundamental question: could AI help people engage more thoughtfully with political issues?
One possibility is that conversational AI encourages people to ask deeper questions about competing ideas, explore trade-offs and examine evidence beyond campaign headlines.
Rather than relying exclusively on short-form political messages, voters can interrogate an issue through multiple questions and request explanations suited to their circumstances.
However, the article also includes an important counterpoint.
Victoria University AI lecturer Dr Andrew Lensen raises concerns that conversational AI can reinforce existing beliefs and introduce less visible forms of bias.
These concerns deserve attention. AI-generated answers can contain inaccuracies, omit important evidence or present information in ways that reflect the limitations and biases of the underlying model.
As Doug Johnstone, CiteCompass Co-Founder, explained to Stuff:
“The value is in helping people interrogate an issue more deeply, not outsourcing the decision.”
We believe that distinction is fundamental.
AI can be useful as a research assistant, but it should not replace independent judgment. Important claims should be checked against original sources, and different perspectives should be examined critically.
The purpose of our study is not to determine how anyone should vote. It is to observe the information environment people may encounter when using these technologies.
Why Independent Media Coverage Matters
The Stuff article represents an important milestone for the CiteCompass research initiative.
Independent journalism creates an opportunity for research findings to be examined outside the organisation producing them.
It can question assumptions, introduce alternative perspectives and bring issues of public interest to a wider audience.
In this instance, Stuff drew directly on CiteCompass research, interviewed members of the research team, reproduced our Week 6 infographic and considered the findings alongside broader questions about the evolution of political communication.
The article also linked readers to the published research results.
We welcome that scrutiny.
The credibility of any emerging research discipline depends not simply on publishing findings, but on making the methodology, evidence and limitations available for independent examination.
That is why we have made our research methodology, weekly reports and interactive data publicly accessible.
Independent Research, Not Political Advocacy
The CiteCompass NZ Election 2026 AI Visibility Monitor was established as an independent, non-partisan observational study.
Its purpose is to measure how political policy information is represented in AI search — not to assess which party has better policies or which political choices voters should make.
We have maintained a consistent research framework across all six monitored parties.
The initiative was reviewed through engagement with the New Zealand Electoral Commission before publication, and the research continues within the original independent, observational scope.
Our guiding principle remains:
Observe. Measure. Report. Do not advocate.
The Research Continues
The Stuff coverage examined findings from the first half of the 12-week programme.
The longitudinal study continues to track how AI visibility changes as the election approaches.
One of the reasons for monitoring over time is that AI-generated answers are dynamic. The sources surfaced, information represented and relative prominence of different organisations can change between reporting periods.
A single snapshot cannot fully explain these movements.
By maintaining a consistent methodology and publishing weekly findings, CiteCompass aims to create a transparent evidence base for understanding this emerging information environment.
We hope the results contribute to a broader conversation about trustworthy information, responsible AI use and the importance of accessible primary sources.
Explore the Research and Media Coverage
Read Finn Hogan’s full article in Stuff
Explore the NZ Election 2026 Interactive Report
About CiteCompass: CiteCompass is an AI search visibility research and analytics platform helping organisations understand how they are discovered, represented and cited across generative AI environments.
Research disclosure: The NZ Election 2026 AI Visibility Monitor is an independent observational study. Its findings do not constitute political endorsements, assessments of policy quality, measures of voter support or predictions of election outcomes.