A Fair-Go for Workers and a Faster AI Future
Why the ACTU’s “right to refuse AI” deserves respect, yet still needs a sharper economic lens
The call for a statutory “right to refuse AI” has reignited an old automation debate: do we slam the brakes or prepare for inevitable disruption? A pure veto right risks becoming both over-broad and under-inclusive. Instead of another hard ban (or a blunt tax), Australia can borrow a smarter policy mix that promotes innovation, rewards novel thinking, and cushions the human fallout.
Australia already struggles with flat labour-productivity growth. Saddling every new scheduling bot or data-entry copilot with a potential workplace veto could:
Slow adoption of benign tools. Firms may shelve low-risk efficiency gains to avoid legal wrangles or antagonism with staff.
Divert investment offshore. Capital is mobile; compliance-heavy regimes push pilots to Singapore, the US or the EU’s lighter “consult first” framework.
In a global race for AI capability, delay costs count.
The ACTU’s stance springs from a legitimate ethical concern: workers should not be guinea pigs for opaque algorithms and have livelihoods put at risk. But a blanket right of refusal risks turning a protective shield into an innovation handbrake.
Here’s what’s happening around the world
Tie public money to joint AI projects that upskill staff
Singapore’s new SkillsFuture Workforce Development Grant reimburses up to 70% of the cost of job-redesign and retraining that accompanies an AI rollout. The catch: a firm only gets the cash if it keeps its existing workers in the pilot.
Canada’s C$300m Compute Access Fund offers discounted GPU time to SMEs, again, conditional on building domestic talent pipelines.
Why this works: it nudges firms toward complementary AI (human + machine) and spreads cutting-edge know-how beyond Big Tech without a single new tax.
Build “flexicurity” into industrial law
Denmark already pairs workforce mobility with portable learning entitlements and mandatory time-tracking to guard against burnout in AI-intensive jobs. A 2024 update tightens those rules for algorithmic scheduling.
Translation for Canberra: make redeployment cheap for employers and automatic for workers by bundling every redundancy with a training voucher or career-transition grant. That keeps labour fluid without freezing innovation.
Insure wage shocks instead of blocking the tech
The U.S. Re-employment Trade Adjustment Assistance program tops up 50% of a displaced worker’s pay cut for two years (cap US$10k). Evidence shows it shortens unemployment spells by roughly three months and lowers political blowback to tech change.
Australia could pilot a similar wage-gap insurance inside the existing Fair Entitlements Guarantee. It would be cheap compared with decades of dole and under-employment.
Make benefits portable for the task-based workforce
A 2025 U.S. Senate white paper finds 80% of independent workers want benefits they can carry from gig to gig. Superannuation is already portable here; consider how sick-leave and learning credits could potentially transition so workers aren’t the collateral damage of enterprise AI deployments.
Put workers inside the design loop, not just at the end of it
The EU AI Act treats recruitment, performance-scoring and other workplace tools as “high-risk,” obliging employers to consult workers or works councils and provide algorithmic impact assessments before deployment. Non-compliance can mean fines up to 7% of global turnover.
Embedding similar co-design duties in Australia’s WHS or Fair Work frameworks would surface bias and usability bugs early, when they’re still cheap to fix.
Attach local-workforce strings to big tech subsidies
The U.S. CHIPS & Science Act folds 33 workforce programs and apprenticeship targets into its semiconductor subsidies. It’s proof that innovation incentives and social safeguards can live in the same statute.
Any Australian quantum or AI-chip package should copy-paste that conditionality.
Subsidise augmentative, not replacement automation
Japan’s industrial policy now channels grants toward human-safe collaborative robots (cobots) that lift productivity without eliminating operators.
Offering an accelerated depreciation schedule for certified cobot deployments would steer capital toward tech that multiplies human skill instead of erasing it.
In summary, smart reform could look like this:
Define “public interest” narrowly. Limit refusal right to clearly high-risk uses (medical triage, automated termination).
Attach skills vouchers to every AI rollout. Make training the default compensation, not litigation.
Offer wage-gap insurance for genuine displacement. US evidence shows a two-year top-up lifts re-employment by 8 to 17 percentage points.
Fast-track dispute resolution inside the FWC so experiments aren’t frozen for months.
Reward augmentative tech. Accelerated depreciation for certified “cobots” that add to headcount productivity.
Better is a co-design covenant:
Workers get voice, skills and safety nets.
Firms get clarity, speed and the green light to chase global opportunity.
Regulate for partnership, not paralysis. Australia can stay both fair and fast in the AI age.