Denny Payne ran a survey of people in Data#3’s contact list to find out about people who are working as AI Engineers in Australia. The data set is a little biased because they must have had some connection with Data#3; the definition of “AI Engineer” is a little vague as well. There were 275 respondents.
Gender Distribution
I’d like to say I’m surprised by this somehow, but I’m not really. It doesn’t look that skewed compared to every other gender distribution in computer science.
Where are AI Engineers located and where did they come from?
Sydney, Melbourne and Brisbane dominate in absolute numbers. No surprises there.
Canberra punches above its weight for AI engineering. But remember this data set doesn’t have public service and defence, so maybe we’re seeing something interesting there. Adelaide and Perth (and maybe Melbourne) are below what you would expect.
We’re dependent on migrants:
And it's mostly India and China as you would expect. Someone thinks Australia is an overseas location; we’ve kept that data point in because you never know what they meant by this.
That’s not that surprising, really. India and China are big countries with huge populations.
When we look at the number of job holders versus population of origin, Vietnam is the place where they are punching above their weight on AI, along with Nepal and Sri Lanka. Not the places I would have expected!
Also a strong showing from Singapore. New Zealand too, but that’s probably more of an indication that labour mobility is a lot easier.
What did AI Engineers study?
Undergraduate engineering degrees are the most common source of AI engineers, followed by computer science. My guess is that this reflects the poor job opportunities for engineers in Australia (outside of mining) over the last decade or two; or maybe that’s lots of software engineers there.
But once we start looking at postgraduate qualifications, it’s mostly a story around IT and computer science.
Where would we be without UTS? (Well probably at UQ, Melbourne or ANU). It makes me sad that Macquarie didn’t make the list. We’re top 50 in the world for data science, and we win awards for cross-disciplinary work!
How long have people been in AI Engineering?
Ignoring people who have been in AI Engineering for a long time (because they don’t fit on the chart), AI engineering is a rapidly growing field with lots of recent entrants.
The number of people working as AI engineers doubles every 26 months. Put another way, if you want to get someone with 2 years’ more experience, the talent pool halves. Or, if trends continue, sometime in the 2060s, the entire Australian population will be working as AI Engineers.
I tried to see if I could see who got into AI Engineering first: UTS graduates, overseas-born, Vietnamese, computer scientists? But it turns out that no, there aren’t any strong predictors of getting into AI engineering a long time in the past. No university turned out to be strongly predictive of early adoption, no location was preference, no degree and neither does gender predict experience. (For you stats nerds out there, if I do a TFIDF model on either the total months, or the log of total months, my cross-validated R2 score stubbornly sat around 0 no matter what I did.)
From which I hypothesise that no-one had any clue that AI Engineering was going to be a thing, nobody planned for it, and people have just lucked into it.