The Intergenerational Report has found its shiny new toy: AI. But what about climate change?

The Seventh Intergenerational Report (IGR) released today by the Treasurer Jim Chalmers looks all the way out to 2066 and yet is mostly focused on the present and the fad-like demand of AI and data centres.

Each iteration of the IGR is less about the projections than its focus. Notionally, Peter Costello established the report in 2002 to focus on the ageing population. It also conveniently enabled him to warn about the dangers of debt and thus justify spending cuts.

As the years have passed, the scope has broadened and the IGR has lengthened. The first was 94 pages long; today’s came in at a rather less svelte 326 pages.

The main issue of each report has been productivity – trying to lift it as it kept falling in reality, and also falling in the estimates within the IGR:

Back in 2002, annual productivity growth of 1.5% was projected, now it is 1.2%, based on a 30-year average.

And if you think such growth is unlikely, that brings us to the topic of the day.

Unlike 2010 which focused on climate change or Joe Hockey’s 2015 version focus on debt, this year it’s Artificial Intelligence – it and ‘AI’ appeared on average 1.5 times per page:

The report early on states “AI will be a defining influence on our economy over the coming decades.”

But the report itself doesn’t even really believe this.

All concrete economic benefits of AI are quite unexceptional. The report notes that “Taiwan, which accounts for more than 90 per cent of global advanced chip production, is experiencing its strongest multi-quarter economic performance since the late 1980s. Similarly, the Republic of Korea has experienced some of its strongest growth in goods exports in 15 years”.

And while that is all fine and dandy, it really is no different from any other economic manufacturing boom – whether it be cars, computers or parts thereof.

For actual benefits, we are left with mostly hopes. For example, the report suggests that “AI also has the potential to make essential services more accessible, affordable and responsive to individual needs.”

Potential is doing a lot of lifting in that sentence.

Similarly, the benefits of having AI data centres built in Australia are rather lacking in explanation.

The report instead takes it as fact that “Australia will also need to ensure competitive access to data, compute and foundational models”. And sure, but does having a foreign company owning a data centre in Australia make it more accessible to other Australian companies?

The report just concludes that “Greater domestic access to computing capacity could support Australian researchers and firms to develop, adapt and deploy AI systems, and may help attract technical talent and AI-enabled businesses”. True it may, but it may also not.

The report even hedges on the economic advantages, noting that “in some cases, hosting training activity in Australia may strengthen Australia’s access to foundational models. In other instances, affordable access to advanced compute at competitive prices may be a more significant factor.”

So yes, it might be an advantage, but if AI is cheaper from overseas, then maybe not.

But the report does note the actual problem of data centres right now, namely that “Data centre construction will also increase competition for capital and resources and add to demands for skills, increasing the importance of addressing capacity constraints across the economy.”

So yes, data centre construction will raise the cost of borrowing, make it harder to get workers for other building works, but hey, check out the potential!

The report is rather lacking in its approach to climate change. Most mentions of the issue come up in the “energy transition” section.

But what is mentioned quite directly contradicts the initial assertion that AI is the key for the economy out to 2066.

The report notes that “over the next 40 years, increasingly frequent and severe natural disasters will pose a significant economic risk. Impacts include the loss of economic activity, infrastructure damage, and higher insurance costs.”

The report also notes climate change will lead to lower agricultural crop yields, increased flooding, erosion and inundation risks for coastal and low-lying areas, and reduced productivity due to reduced pace of work and more breaks due to heat.

Moreover, the report suggests that if global temperature rises to 3C above the pre-industrial average rather than the 2C aim of the Paris Agreement, GDP by 2066 will be 1% lower:

Unfortunately, the government remains committed in the IGR to the Safeguard Mechanism to “continue to drive decarbonisation in Australia’s largest industrial facilities.”

There is alas no mention that the current pace of emissions reduction under the Safeguard Mechanism is woefully inadequate:

The report notes the increased needs of the community for government services. For example, the cost of the government’s healthcare expenditure is expected to double from 3% of GDP to 6% of GDP:

The IGR however, just assumes government tax revenue will remain steady at 24.2%. And with no tax changes, personal income tax will reach 14.1% of GDP:

But revenue from the gas industry continues to elude. The report notes that “PRRT receipts are assumed to remain constant as a share of GDP beyond the forward estimates”. Given that a 25% tax on gas exports would raise $17bn a year – or just over 0.5% of GDP, the IGR makes the case for an export gas tax, even if it fails to state it.

The Intergenerational Report, despite being pitched as a long-term document, is, as inevitably is always the case, about the current moment. This was the AI Report – blinded by the bright, shiny new toy.