Daily AI Brief

The World Trade Organization's 2025 World Trade Report warns AI could raise the value of global trade in goods and services by 34-37% by 2040 and boost global GDP by 12-13%, if policy and technology gaps are addressed. Key enablers: reducing trade costs, improving productivity, better digital infrastructure, workforce education and inclusive regulatory frameworks. Without those, poorer and middle-income countries may benefit much less, risking a widening economic divide.

A version of DeepMind's Gemini 2.5 solved a complex real-world programming challenge in an international competition, beating human coders on a task involving routing fluid through interconnected reservoirs to optimise speed. Although it failed two out of twelve tasks, its overall performance placed it second out of 139 top coders. DeepMind claims this is a "historic" advance in abstract reasoning.

According to Gartner, global spending on artificial intelligence will hit about USD 1.5 trillion in 2025, driven by infrastructure expansion (data centres, GPUs), integration into devices (smartphones, PCs), and adoption of AI cloud and edge services. The report predicts the market will exceed USD 2 trillion in 2026. Growth is not only in traditional tech hubs; newer cloud / AI-service providers and regions are gaining momentum.

Nvidia, OpenAI and infrastructure firms like Nscale and CoreWeave are collaborating to expand AI infrastructure globally, with large-scale deployment of GPUs, data centre build-outs, and projects intended to give regions more sovereign control over their AI compute capabilities. The UK is a major focal point, but multiple regions are part of this push.

Researchers in New Zealand have prototyped a decision-support tool for homeowners that incorporates anomaly detection, baseline modelling, and scenario simulation (for insulation, LED retrofits, etc.), pulling together main relevant datasets. Experts rated its usability highly. It aims to help translate national policy and subsidy programmes into practical, personalised guidance.

The WTO report strongly argues that unless lower-income countries, and by extension less developed regions, close the technology, regulatory, and infrastructure gap, the promise of AI will worsen inequality. Emphasis on predictable trade rules, lower tariff barriers for AI-relevant hardware, investment in workforce training.

NZ researchers and experts are calling for good governance practices to be built into AI deployment; this includes transparency, accountability to communities, fairness, inclusion, as well as ethical frameworks. Such calls often accompany new tools, government programmes, or local regulation consideration.

DeepMind's model solved complex tasks in less time than many human competitors could, especially for routing/distribution problems involving unlimited search space or many variables. This suggests progress towards more general reasoning rather than narrow domain performance.

The tool for energy-efficiency in residences is being evaluated by domain experts; its modular dashboard approach may be replicable for other policy-driven tools.

"AI could boost trade by nearly 40% but tech and policy gaps remain, WTO finds" — deep dive into what's needed so economic gains from AI are spread broadly.

"Google DeepMind claims 'historic' AI breakthrough in problem solving" — illustrates progress in reasoning and abstract problem solving in AI models.

"AI spending to reach $1.5 trillion this year, Gartner says" — good context on scale and pace of investment.

Global: WTO's warning that inequality may grow without policy catch-ups; infrastructure and regulatory regimes are central.

NZ: Local experts pushing for strong governance, especially as more AI deployment is expected in public and private sectors.

The energy-efficiency decision support tool in NZ: if adopted or scaled, could serve as a model for how government-backed AI tools can bridge policy and household level impact, particularly in climate and health sectors.