The AI ‘beginner tax’ worth billions

According to new research, the economic and productivity potential of AI remains untapped in Australia due to gaps in capability, judgement, and management.

The findings also reveal how and why Australian businesses should capitalise on the opportunity by upskilling, training, and preparing staff now, and potentially reconsidering who is making the biggest decisions.

A new report from RMIT Online and Deloitte Access Economics has found that more than 5 million Australian workers are still showing beginner-level AI literacy – and it’s reportedly holding back employees and organisations. Despite 84 per cent of polled Australian workers using AI, 54 per cent exhibit only low-level competence.

According to the study, if half of this proportion improved to even intermediate-level AI competence, the wage and productivity benefit alone would exceed $18.9 billion annually. For employees, this could mean about $7,000 extra in annual wages for full-time workers.

Progressing from beginner to advanced literacy would be worth over $10,000 annually.

However, the report highlighted that workers must be able to use AI safely and strategically in order to create meaningful value. As such, one factor to address is how AI is used according to generation.

Understandably, Boomers are more likely to remain at beginner-competency levels than younger workers, with over three-quarters possessing basic competence, compared to 43 per cent for Millennials. If Boomers were to come closer to this proportion, it would reportedly unlock over $3 billion.

The crux of the issue, however, is not merely economic, but managerial. As would be expected, the more experienced and older employees tend to fill higher, more senior and often decision-making roles. Their understanding of and skill level with AI, therefore, is guiding organisational deployment.

Differences in AI competency aren’t the only factor at play. As the report highlights, certain skills different generations bring to the AI use are creating a dangerous imbalance. As it is, employees are twice as likely to display advanced technical AI skills, such as prompting, than they are in critical judgement capabilities such as ethical awareness, risk assessment, and critical evaluation of outputs.

And then the generational divide compounds the issue: Millennials are more likely to demonstrate stronger technical capability but overestimate their literacy, thereby deploying AI without adequate oversight, while older workers are both more hesitant to adopt AI and not equipped with the technical knowledge to conduct such oversight.

In addition, the report explained that technical skills can often be self-taught through experimentation, whereas judgement skills develop more slowly and usually require deeper domain expertise and training to be able to conduct ethical reasoning and assessment of bias, risk, and unintended consequences.

As such, businesses are facing the dual challenge of ensuring younger employees develop greater judgement skills and supporting senior leaders to build the confidence and capability required to lead adoption.

And in the meantime, organisations are seeing systemic errors. Previous research showed 95 per cent of companies reported experiencing an AI-related incident over the past two years, suffering an average financial loss of $800,000, and more than 50 per cent of polled workers have made a mistake associated with AI use.

More than half also admitted to passing off AI-generated content as their own.

While the report found that AI is saving the average worker nine hours a week, 47 per cent of these employees are simply reallocating that time to completing other tasks, and only one-quarter are furthering skill development. According to the associate director at Deloitte Access Economics, Rhiannon Yetsenga, “The productivity gains from AI are undeniable, but the focus now needs to shift from mere adoption to competency.”

“The real opportunity is equipping workers to use AI effectively, strategically and ethically.”

RMIT Online CEO Nic Cola said: “The research makes it clear that self-guided, ad-hoc experimentation is not enough to move the needle on national productivity. We are seeing a landscape of ‘shadow AI’ where nearly half (49 per cent) of workers are teaching themselves through trial and error, often building surface-level technical skills while critical judgement lags.”

The importance of embedding AI literacy through structured training, and not leaving employees to informal self-teaching, is clear. This, the report urged, is an investment for both employers and employees, and also protects organisations from legal or reputational incidents.

And with more than half of workers not receiving any AI training from their employer, and only 11 per cent receiving structured and ongoing support, a meaningful lift in national productivity will only be seen if employees at all levels are better equipped to use AI responsibly and effectively.

As the report surmises, businesses should target critical evaluation and transferability skills, prioritise practice use cases, provide clear rules on permitted AI use, and tailor training to different levels of experience and confidence.

Cola said: “For employees and the economy to benefit, employers must move away from informal guidance and invest in structured, accredited learning that prioritises critical thinking and strategic application.”