Te Pāti Māori: can they land the waka?

And can AI eliminate corruption and cheating, rather than make it worse?

The maiden speech by Te Pāti Māori’s new MP, Oriini Kaipara, was the perfect time to announce the party’s strategic “reset”. TPM has fought “the most racist and fascist government we’ve seen in a generation”, they said, and it’s time to set a new course on behalf of their people. They aim to make Luxon & Co a one-term government. And they want to make a difference for their people in a future coalition with Labour.

TPM is doing relatively well in opinion polls, within range of the five percent MMP threshold. As in the lead-up to the last two elections, it’s possible that polls are still slightly under-estimating TPM’s support. At this rate, they could increase their number of seats.

They want to be in a position to “choose” who the next prime minister will be. Clearly they won’t choose Christopher Luxon – so one can only suppose that they’ll have to choose Chris Hipkins.

Without making predictions, let’s imagine, then, that Luxon concedes defeat after the next election, as his three-party coalition has lost its majority. What scenarios could TPM face?

A not-so likely possibility is that Labour and the Greens have the numbers to form a two-party majority coalition, and TPM’s seats are surplus to requirements. There could still be a minor role for TPM – a role they might not want to play under those circumstances.

But if Labour plus Greens add up to less than a majority, TPM could then be needed as either a coalition partner, or a confidence-and-supply support party. Some of their recent statements have made it sound as if Labour and TPM wouldn’t get along well together, but time can heal that – which apparently was one of the reasons for the “reset”.

Normally, the baubles of office motivate politicians to be on their best behaviour. But the continuing spat between TPM and Labour MPs, and the fallout from the Tāmakai Makaurau by-election, don’t show a lot of healing at this stage.

It’s not unthinkable, moreover, that NZ First could do a deal with Labour and cut TPM out. Remember the Labour/NZ First coalition supported by the Greens in the 2017–2020 term? Winston Peters could explain transitioning back from National to Labour as rescuing the country from TPM’s radicalism and chaos. Stranger things have happened in NZ politics: TPM once supported a National-led government, for instance.

So TPM need to prove that they can be a credible partner in a future left-wing government, or they risk being left out. Opponents will claim that TPM’s radical policies and their propensity to disrupt parliament’s tikanga argue against any change of government at all in 2026.

Furthermore, TPM have their internal problems.

In a media stand-up, TVNZ’s Maiki Sherman gave co-leaders Rawiri Waititi and Debbie Ngarewa-Packer a chance to talk about their leadership style by recalling an accusation that the party was run like a dictatorship. It was a golden opportunity for TPM’s leaders to say what they really stand for. Instead, Waititi called the interview to a close and, taking Ngarewa-Packer by the arm, turned and left.

An unsuspecting viewer may have been left wondering if the accusation of dictatorial leadership had touched a nerve because it was true. Or are TPM unable to cope with the pressures of real-world political leadership?

They’ve been taking a strict line with the media, and so there was a soft-ball interview with Mahingarangi Forbes.

If a reporter wants to interview TPM leaders, it needs to be on their terms: no hostile questions, no racist assumptions. As a “protective” media strategy, it could work for them, however, by showing their support-base that they’re not being pushed around by the forces of colonialism, even if it costs them some air-time.

The pointed question about dictatorial leadership was a breach of trust and was disrespectful, they told Forbes, as it distracted from that day’s celebration of “the reset of this waka”. And, by way of substantial policy, they want more public procurement to go to Māori contractors and workers, for example, and they want to see a constitutional transformation.

Anecdotally, some people cite TPM as a reason not to vote Labour. If office-holding is their aim, they need to address that fear factor. Back in February, they announced that a bottom line for any coalition agreement would be a new Parliamentary Commissioner with power to veto bills that don’t comply with Te Tiriti o Waitangi. (A Parliamentary Commissioner is normally appointed, not elected, FYI.) That’s a constitutional reform of huge significance – indeed, significant enough to necessitate a referendum, as a veto could override parliament’s supreme lawmaking power. Is New Zealand up for another divisive debate?

Update: This column was written before news broke about alleged overspending by the office a TPM member.

---

On bureaucrats and intelligence – human and artificial

I’m currently reading anecdotes and memoirs written by a retired civil official, Lu Rong (1436–1494), who had served under China’s Ming dynasty. (You can read, for free, selections from his Miscellaneous Records from the Bean Garden.) To put it in perspective, in the year that Lu Rong died, the French invaded the Italian peninsula, precipitating the banishment of the Medici family from Florence, and Henry VII was king of England.

The Chinese had a centuries long tradition of scholar-officials who underwent competitive ordeals of classical learning and invigilated examinations to qualify for highly prestigious roles in the civil service. Dynasties rose and fell, but the bureaucracy lived on. Confucian moral integrity and benevolence were of the utmost importance to these officials, but they were only human. What we would now call “corruption” was rife, in large part because they weren’t paid enough. At that time, though, they were the world’s largest educated workforce.

Zhu Yuanzhang (1328–98), the first Ming emperor, or Hongwu Emperor, oversaw the execution of 30,000 to 40,000 people in his efforts to clean out the previous Mongol Yuan dynasty’s bureaucracy. (The Trump administration’s sackings pale in comparison.) But, over time, the bureaucracy reasserted itself, and later Ming emperors were prevented from taking initiatives. Ritual, memo-writing and official connivance prevailed.

Today, however, China is forging ahead with AI.

Two AI-related themes in the media recently are: it will eliminate jobs, and it’s being used to cheat at university. But can AI be used constructively in public administration, and in particular to clean out corruption?

Albania’s Prime Minister Edi Rama has (rather theatrically) announced that he’s “appointed” an AI, affectionately known as Diella (or Sun), to his cabinet. Its “portfolio” is the elimination of corruption from public procurement. It’s a way of saying that the cabinet is being watched by a non-corruptible “clean” machine – but, as always, a lot will depend on the laws and rules on which the AI is trained, and whether humans will follow up and prosecute any breaches that it flags.

Transparency International’s 2024 corruption perceptions index ranked Albania 80th out of 180 countries, but its score had improved since 2021. Let’s hope that Diella helps them make further improvements.

The OECD has published a paper on governing with AI. The most frequently cited reasons for using AI in government have been automation, streamlining and tailoring of processes. But, in procurement and fighting corruption, AI is being used to enhance accountability and detect anomalies. In designing such systems, though, it’s important that front-line public servants and end-user citizens have some involvement and influence. It’s also essential to have “pause” switches, where the AI seeks human oversight and judgement before moving to the next step, let alone making any decisions that affect people’s lives.

But the danger is that, even if AI flags an anomaly in a tendering process, the humans fail to take any action on it – or the rules weren’t good enough to begin with – and so AI-monitored procurement, such as in Albania, may then be just a cover-up.

“We’ve built a machine to keep ourselves clean.” That belongs on a Tui billboard. AI can be used to cheat and to deceive, after all.

Some universities are distinguishing themselves as “dumb” adopters of AI. It’s turned out that using AI to detect undeclared AI-generated writing isn’t very effective. The Australian Catholic University hit the news after accusing 6,000 students of academic misconduct, forcing many to prove their own innocence. The trouble may be that some students write like robots because they use AI to improve their writing (but not to do their whole assignment), or conversely that AI is capable of writing like an undergraduate (if you ask it to), or that clever students deliberately mess up AI-generated writing to make it look authentically human.

RNZ’s John Gerritsen reports that several New Zealand universities “have given up using software to detect the use of Artificial Intelligence in student work”.

The problem arises as academics try to punish AI use, rather than adopt AI as a learning tool, using it in ethically responsible and intellectually constructive ways.

AI may mess with our minds, if we let it. Readers are no doubt well aware (ad nauseam) of the problems of misinformation and disinformation. What’s often overlooked are good old facts, truths and reasonable opinions (which still exist) and the changes in how we’re learning about them. The internet, social media and now AI are transforming the means by which people form their political opinions, when based on truthful or verifiable information. AI will have the ability to spread valid facts and mainstream political discourse at scales and speeds never seen before, and with messages and images individually tailored for each of us. In a free and fair system, this will make it more likely that you, the consumer, will be nudged one way or the other, even if you’re a critical and discerning reader/viewer who’s not prone to conspiracy theories.

McLuhan’s maxim “the medium is the message” remains relevant: impressions and opinions are formed in a social, institutional, technological and communicative context – not by individuals rationally distinguishing right from wrong in splendid Cartesian isolation with no internet connection. The institutions, norms and practices of opinion-formation are changing rapidly – and it’s only getting faster, weirder and more confusing. AI will change and accelerate the ways in which we gain impressions that “most others agree/disagree” about something important to us.

Even reasonable, evidence-based discourse will shape opinions in ways we hadn’t dreamed of before. As fair interpretations of facts and evidence can differ markedly in moral and political terms, this effect of “good” AI will be polarising and confusing – even if we could avoid outright lies and fakes.

What we’ve already learned from the “news finds me” attitude of social media users is that education systems will need to train people to be active readers and researchers rather than passive consumers in the coming super-saturated information environment.

I promise not to train up a GrantBot to write my columns while I sip my latte.

The emperor (in yellow) receives an examination script. The Song Dynasty’s finest.