AI usage has accelerated, but remains relatively low in most countries
One way of tracking the potential usage of AI is to rely on the Indeed Hiring Labs measure of job ads that include a reference to artificial intelligence as a share of all job ads, where central banks monitor Indeed data on the labour market.
On this metric, usage of artificial intelligence is still relatively low in most advanced economies, ranging between 4% and 7% of all ads, except for Canada, where ads that refer to AI account for 13% of ads.
Judging the take-up of AI on this measure, ads have recently accelerated relative to simple trends that were estimated prior to the public release of ChatGPT in 2022 in a few countries, namely, Canada, the UK and Australia, but, surprisingly, remain broadly in line with pre-ChatGPT trends in the euro area and the US.
Interestingly, the take-up of AI – which is skewed towards large companies – seems narrowly based at this point in that the Australian branch of Indeed reports that, “almost two-thirds of AI-related postings came from the top 1% of firms by AI posting volume, a share largely unchanged over the past two years”.
The occupations in Australia with the highest share of ads mentioning AI are skewed towards tech, namely, software development (43%), data and analytics (43%), IT systems and solutions (27%), industrial engineering (18%), and marketing (17%), where adoption of AI is “still primarily driven by a relatively small group of employers”. Banking and finance is down the list of occupations, but at 13% it has a greater share of ads than the economy-wide average of 6%.
In terms of whether AI is displacing jobs, Indeed reports that in Australia to date, “overall hiring trends show little sign of disruption”. Ads for the occupations with the “highest exposure to AI tools”, which includes “accounting, marketing, administrative assistance, and banking and finance”, has remained broadly stable at about 30% of job postings, suggesting that “recent job posting trends … reflect the broader economic cycle rather than widespread AI-related displacements”.
4 topics