AI needs a steady supply of good data to work – and data experts are a key part of building that supply
One of the fastest-growing industries in the world today is AI. Machine learning spans dozens of different fields – you might be most familiar with large language models, but it’s also a vital tool in everything from weather prediction to medical image analysis.
Whatever the application, AI systems need huge amounts of data to learn. Early versions of ChatGPT were trained on about 300 billion words!
AI and machine learning are also important for processing huge data sets. For example, CSIRO’s ASKAP radio telescope generates so much data that astronomers use machine learning to classify it and pull out interesting info.
If this data is biased or faulty, it produces a biased and faulty AI. So it’s extremely important that data scientists understand and manage the data going into AI training. There’s a lot of important work in training AI models, and checking
Unlocking local data sets
If we want AI that’s useful to Australians, we need to train it on Australian data. In its National AI Plan, the government highlighted the importance of Australia’s high-quality data sets in the public and private sectors for supporting local AI.
It’s also important that people maintain control of their data – particularly for Aboriginal and Torres Strait Islander communities. Data scientists will need to take data sovereignty seriously while developing AI.
Putting on the brakes
What if you have sensitive info that you don’t want an AI to absorb? There’s actually a lot of work in cyber security focused on preventing AI systems from accessing certain data.
This includes building image filters that stop AIs from learning on certain images in order to protect people’s privacy and intellectual property. Cyber security experts have been working on filters that do this.
They’re also helping organisations protect confidential data, such as health records and identifying documents. It’s important to have people who know how to keep this stuff safe!
Big (giga)bytes
- An hour of Netflix in high definition: 3GB
- All of the text on the English language Wikipedia: 25GB
- Data produced by the ASKAP radio telescope in 1 second: 12,500GB (12.5TB)
Start your career here
Data Science + AI + Study
- Bachelor of Data Science (Artificial Intelligence and Machine Learning), QUT
- Bachelor of Computer Science (Artificial Intelligence and Machine Learning), Adelaide University
- Master of Data Science and Artificial Intelligence, University of Newcastle
- Diploma of Artificial Intelligence, TAFE Queensland
Data Science + AI + Jobs
- AI specialist, $52K–$257K
- Data scientist, $69K–$148K
- Machine learning engineer, $62K–$149K
This article was first published in Careers with STEM: Data Science, produced in partnership with QUT.
