With most businesses now embracing generative AI (genAI) for specific use cases, limitations on the availability of skilled AI practitioners are forcing many to turn in-house – but experts warn that without a clear understanding of what skills it actually takes to pull off a genAI implementation, or how to get them, many companies are flirting with disaster.
Although employers believe AI could boost productivity by 51 per cent if integrated into their operations, the recent AWS Study on AI Skills in APAC found that 75 per cent can’t find the AI talent they need – and that 79 per cent don’t know how to train their workforce to build that talent in-house.
“The area is changing rapidly, and while companies recognise that they need to be AI ready, they don’t really know where to start,” explained Leif Pedersen, APAC cloud and AI product manager with training firm Lumify Work, the recently rebranded company formerly known as Dimension Data Learning Services (DDLS).
Lumify Work offers a range of AI-intensive training courses, including its 8-week CloudUp and AWS Generative AI Accelerator boot camps, AWS Skill Builder on-demand digital training, and a range of AI-related practitioner certifications.
“A lot of what we’re seeing in the Australian market is just ‘where do I start?’ and ‘how do I understand what this could mean to my business?’,” he continued.
“While there are some organisations doing some pretty cool stuff, the vast majority are quite new to the technology. It’s very much toe-in-the-water in this region – and it’s changing very rapidly, so it’s hard to keep up.”
An AI skills roll call
For companies just moving into the AI space, managers believing that AI capabilities are just something that existing technologists and developers can pick up, are bound to be disappointed.
Gartner, for one, recently surveyed over 700 companies in the US, UK, and Germany and found that 87 per cent were already operating dedicated AI teams – and that two-thirds were creating new roles specifically for genAI.
“The deployment of AI and genAI within organisations is demanding specific capabilities, “ Gartner VP analyst Jorg Heizenberg said, “meaning that for the data and analytics team, the investment in new talent is real, urgent and required.”
“There is no benchmark that determines the size of the D&A team as there are too many dependencies. The question that chief data analytics officers [CDAOs] should ask themselves is how many roles they need to ensure their teams are successful and effective.”
The true breadth of AI-related roles will surprise many: contemporary AI teams must, for example, include essential roles such as machine learning engineers, data engineers, data scientists, prompt engineers, AI ethicists, and translators between data and analytics and AI.
Project teams require strategic heads that can develop an AI strategy and shepherd it through its lifecycle, as well as business needs analysis and development capabilities such as AI architect, AI risk and governance specialist, AI product manager, analytics engineer, UX designer, and AI developer.
With genAI expanding the reach of AI to sensitive customer-facing interactions – where issues like privacy and ethics come into play – companies must also consider the importance of emerging AI roles like knowledge engineer, model validator, and decision engineer.
It is also crucial to bring everyday workers into this change process, Gartner warned, with 39 per cent of companies running genAI literacy programs to help workers consume, analyse, and make smart decisions with data – an essential part of the transition since new AI tools are ultimately being implemented to benefit those workers.
Companies that fail to invest in their workforce’s AI skills now will regret it down the road, Heizenberg warned, with a quarter of staff attrition expected to be due to managers’ lack of data literacy.
“Business leaders should not ignore data and AI literacies,” he said. “They are interrelated as half of AI techniques are fuelled by data, and data literacy is essential to AI literacy and vice versa.”
Bootstrapping the internal AI team
Rather than trying to build a comprehensive roster of AI skills from the get-go, AWS technical trainer Peter Vandaele recommends that companies “choose a use case and start simple.”
To build developers’ AI capabilities, Vandaele said, developers might start working with the AWS Bedrock AI foundation, then work with the “amazing foundational models” – including human speech and natural language processing – to grow into ever more complex and business-driven applications.
“If you start small,” he said, “you can get started pretty easily with stuff like Bedrock.”
“There’s a whole bunch of capabilities in there that really help you go through this whole MLOps journey, from setting up and preparing your data to setting up the infrastructure, to actually training your model.”
Steadily evolving AI use cases will also build companies’ understanding of their skills requirements, allowing them to target further training to deliver the biggest benefit to the business – and the most effective assistance to expanding AI teams.
“It really needs to come back to what you are trying to achieve by using this technology in your organisation,” Pedersen said. “Do you just need to understand what AI is and how it can improve your processes or help your people?”
“You can get very technical very quickly,” he added, “as you move to building really sophisticated AI models and engines, ingesting huge amounts of data and coming up against issues like bias, ethics, and AI hallucinations.”
“If there’s a use case out there, there’s a fair chance we’ve delivered training around it in some way, shape, or form. Ultimately, you can be as technologically advanced as you want to be or need to be.”
Click HERE to learn more.