It was not long ago that artificial intelligence (AI) was solely the subject of science fiction movies.
Subscribe now for unlimited access.
or signup to continue reading
Fast forward to 2024, and 50 per cent of lawyers are using it to write briefs, doctors are using it to diagnose conditions, programmers are using it to write their code and airlines are using it to improve safety and efficiency.
Like any new technology, AI will have impacts on the workforce that need to be managed. But, it turns out, those impacts won't be felt evenly across the community. Some will feel it more than others.
To find out who will bear the brunt of AI - specifically generative AI (think: ChatGPT) - our team at Mandala used workforce and careers data to classify jobs into three categories: insulated, augmented and disrupted.
"Insulated jobs" are those that have a small portion of skills that are replicable by generative AI.
"Augmented jobs" are those where a large portion of the core skills are both replicable by AI (i.e. replaced by the technology) and complementary to generative AI (i.e. allowing the worker to focus on other, higher value tasks).
"Disrupted jobs" are the most problematic. These are the jobs where a large share of the core skills are replicable by generative AI and a small share of skills that are complementary to generative AI.
In short, those in insulated jobs will be largely unaffected, those in augmented jobs will see the nature of their jobs change, and those in disrupted jobs will need to retain, reskill and potentially find new jobs.
What do the results look like?
We found that about 7.1 million workers - about 50 per cent of the workforce - are in the insulated category. They will be relatively unaffected: everyone from dentists, doctors and drivers through to vets, farmers and real estate agents.
We found that about 3.3 million workers (23 per cent of the workforce) fell into the augmented category, meaning that their skills are predominantly complemented by generative AI and would use their existing skills to shift their focus to higher value-add tasks.
This category includes everything from software engineers, data analysts and web designers through to cashiers and marketing managers.
Finally, about 3.9 million workers (27 per cent of the workforce) fell into the disrupted category, meaning their skills are predominantly replicable by generative AI. This means that, overtime, these workers may need to reskill and focus on alternative jobs.
These occupations include legal associates, finance officers, administrative assistants, translators and librarians.
When we analysed the occupations most impacted by generative AI and match this with data on the characteristics of these workers, we start to get an idea of how generative AI will impact the community.
These impacts, it turns out, are not shared evenly.
The occupations most impacted by generative AI are those which disproportionately hire women and young people.
Unlike previous episodes of technological change, it will also disproportionately impact occupations in services than other sectors like manufacturing.
Generative AI will also impact occupations with higher education requirements, another stark difference to previous bouts of technological disruption.
So, what should we do about this?
The key value of these datasets is that they let us target supports to those who need it and tailor those supports in the right way.
Those who are insulated from the impacts of generative AI shouldn't receive any supports. It sounds obvious, but policymakers often get this wrong.
Some European countries in the climate transition, for example, have retired coal miners and paid them their salary for the rest of their lives.
It's generous, but it's bad policy because it means lots of skills that are in high demand (e.g. electricians) are taken out of the labour market.
For generative AI, supports should be targeted at the second two groups: those whose jobs will change and those who will need retraining and reskilling.
For those whose jobs will change, the onus will be on organisations to help their workers to refocus their efforts on the areas that are complementary to the tasks that AI will replace.
For those workers who will need to retrain and reskill, the onus will be on governments, and there's lots they can do.
The current patchwork of retraining and reskilling programs across states and territories should be brought together by the federal government and made more generous.
READ MORE:
Payments to unemployed people and people on youth allowance should be increased.
Scrapping non-compete clauses will help people switch jobs.
Naming and shaming gender pay gaps is a step in the right direction, but without addressing things like paid parental leave, affordable and accessible childcare and social expectations around care giving roles, the gender pay gap will likely persist unless the government acts in these areas.
The big risk is that the impact of AI will worsen the disadvantage already faced by women and young people. These are fault lines in our society which, when faced with shocks like technological change, will worsen and deepen.
The government shouldn't need any more reasons to address the challenges facing women and young people. But if they do need another reason, generative AI just gave them one.
- Adam Triggs is a partner at the economics advisory firm, Mandala, a visiting fellow at the ANU Crawford School and a non-resident fellow at the Brookings Institution.
