Leaning into AI: Keynote speakers

As AI becomes embedded in our everyday lives, organisations are harnessing its power and affordability to revolutionise the way we do business. Across two Leaning into AI seminars, SmartCompany and AWS partnered to explore the rise of AI, the current state of play and how businesses can make the most of the technology. 

Among the esteemed guests, we heard from three keynote speakers: 

• Simon Johnston, AWS Artificial Intelligence and Machine Learning Practice Lead for ANZ
• Augustinus Nalwan, Carsales GM of AI, Data Science and Data Platform
• Mike Bewley, Nearmap Senior Director, AI Systems

Let’s recap what these experts had to share.

Simon Johnston — AI services for the mainstream

Simon Johnston made one point very clear: AI and ML tech is here, and it’s good to go. “60% of businesses are thinking about implementing AI in at least one part of their organisation,” says Johnston. “At AWS, we have over 100,000 customers already using it. It’s no longer just experimentation, it’s a mainstream endeavour that businesses today are taking on board.”

Johnston’s point was the seminar’s biggest takeaway: AI tech is ready for all businesses, regardless of data capabilities or buying power. At AWS, the technology is separated into three levels of approachability and purpose — the ‘tech stack’. 

At the bottom is computing power provided for machine learning builders. The middle — AWS Sagemaker — provides data labelling, model building and model tuning capabilities, while the top layer is about services that are accessible to every business, not just those with data scientists on staff. The point is this: however your business dreams of using AI, it’s now possible. “The bars for entry, I believe, are gone,” says Johnston. “It’s very cost-effective to test these technologies out to see how they work within your business.”

Simon Johnston — The trends

The other major element of Johnston’s keynote was breaking down the current trends in AI and ML. If you’re looking to upskill yourself on AI tech (and particularly if you want to use it for business success), it’s worth understanding the following:

• Model sophistication: ‘Foundation models’ are the broad datasets used to train machine learning, and they can be applied across a wide spectrum of uses. With access to complex, pre-trained models, we’re seeing businesses adopt innovative solutions like Stable Diffusion from stability.ai for image generation.
• Data growth: Businesses have a huge amount of data, but it’s not always usable. Tapping into ‘unstructured’ data (the type that is typically difficult to use, such as image files) with tools like Sagemaker Ground Truth is helping businesses take advantage of AI.
• ML industrialisation: Machine learning on an industrial scale means creating, testing and deploying models at a great speed through unlimited growth, as in Johnston’s AstraZeneca example. “It used to take them three months to put a model into production – it now takes them one day.”
• ML-powered use cases: Use cases are exploding, with approachable solutions across the business spectrum, including intelligent document processing, inventory management, fraud detection and data labelling.
• Responsible AI: AI can’t just be about unchecked power — it has to have a positive benefit. To do that, Johnston points to six markers of responsibility: fairness, explainability, robustness, privacy/security, governance and transparency.
• Democratisation: This is one of the biggest trends in AI and it’s all about bringing the technology to as many users as possible. This happens through education and tools like Sagemaker Canvas which don’t require a data science degree to use.

Mike Bewley and Augustinus Nalwan — AI and ML in practice

While Simon Johnston gave us a high-level look at AI, Mike Bewley and Augustinus Nalwan provided some more personal insights. 

Bewley’s business Nearmap uses machine learning to build detail into geospatial imagery for organisations like insurance companies and emergency services, distinguishing between different types of debris, levels of building decay and vegetation. In order to focus on its business goals, Nearmap has found success in outsourcing some of the more routine ML processes. “Outsourcing generic problems is really important,” says Bewley. “We’re not a huge company that can afford to build huge platforms for everything we do — you’ve got to leverage it. If someone else can do it, you can focus on the problems that matter to you.”

At Carsales, Augustinus Nalwan found success with AI solutions like car model recognition and automatic blurring of licence plates. As the business has come to support a more AI-led business model, Nalwan has had to make a choice: hire more data scientists and ML experts, or upskill the staff he already has. By choosing the latter, Nalwan has been an early adopter of AI democratisation. Leaning on Metaflow and Sagemaker ecosystems, Carsales have been able to automate much of the ML workflow, bypassing the strict need for data scientists. “What we’re doing here is we’re turning our software developer into a citizen data scientist and citizen ML engineer,” Nalwan says.

Learn about the 6 key trends driving Machine Learning innovation across Australian and New Zealand industries inclusive of improvements to Model Sophistication, Data Growth, ML Industrialisation, ML Powered Use Cases, Responsible AI and ML democratisation.

On-Demand Keynote Recording: View Here

Read now: Hear from the experts: using machine learning and AI to grow your business