SUMMARY

Artificial intelligence today sits at the crossroads of grand hopes and rampant hype. Optimists point to extraordinary investments—tech giants like Microsoft, Amazon, Meta, and Alphabet pumping hundreds of billions of dollars into AI infrastructure—as catalysts for growth and efficiency. In the U.S., AI-driven capital expenditures in the second quarter of this year accounted for nearly half of that quarter’s 3% GDP gain. McKinsey forecasts an astonishing $7 trillion in global AI investment over the next 5 years, fueling narratives of transformative potential.

This wave of optimism has also reached Australia. Political leaders—including Treasurer Jim Chalmers— note “We see AI as a huge opportunity for Australia, it’s a key part of our productivity agenda, an absolute game changer”, aligning with Productivity Commission projections of AU$116 billion in gains over the next decade and a 4.3 percentage‑point boost in labour productivity. This marks a significant shift from the earlier priority on safe, ethical deployment to one of economic opportunity.

This shift extends beyond Australia. In the United Kingdom, the new Labour government’s AI Opportunities Action Plan—backed by £14 billion in private-sector investment—marks a clear pivot from previous safety-centred messaging to one emphasising “opportunity and growth.” However, history offers a cautionary note—productivity gains are rarely distributed evenly. A recent Nature paper, Bridging the Digital Divide, examined data from 59 countries over 25 years and found that while technological innovation can enhance connectivity and efficiency, it has often exacerbated income inequality, especially in developed economies.

And then there is the question of whether these forecast productivity gains are realistic. Many economists urge caution. Torsten Sløk, Apollo’s chief economist, warns the AI bubble may rival the 1990s dot‑com boom, while Nobel Laureate Daron Acemoglu estimates AI may automate only 5% of jobs and contribute approximately 1% to global GDP this decade, a non-trivial but modest effect, and certainly much less than the revolutionary changes some are predicting. Emily M. Bender’s The AI Con also challenges techno‑optimism by characterising many AI systems as “sophisticated illusions” that mimic understanding without genuine comprehension, underpinned by invisible labour and structural inequality.

A more balanced view is given by Narayanan and Kapoor, in AI as Normal Technology, urging us to understand AI as akin to electricity or the internet—transformative but gradual, institutionally mediated—not as a sudden rupture, tempering both utopian and apocalyptic visions.

This evolving discourse on AI’s impact is at the heart of the upcoming Royal Society of NSW Forum Day: “AI: The Hope and The Hype.” The event’s sessions, AI in Communities, AI for Health, AI and the Law, AI in Practice, and AI Research & Future Directions, offer a timely platform to explore how AI’s promise intersects with productivity, equity, governance, and societal impact.

At its core, the tension remains between hype — exuberance untethered from evidence, and hope — grounded optimism shaped by governance, ethics, and fairness. As we confront this new technology, a critical question emerges: Is AI truly different from previous transformative technologies—like electricity or the internet—or is it simply another innovation riding waves of hype?