Column: Tackling inflation in the era of generative AI

The biggest increase in Australian inflation in 32 years has well-known causes. A global pandemic affected supply chain dynamics. The buying public increased demand for goods to support a hybrid work environment.

Energy costs escalated in the wake of an atrocious outbreak of war and in response economic stimulus switched to tightening. The humble act of purchasing toilet paper became a metaphor for the challenges many organisations faced.

The value of a strong supply chain, and reliable suppliers founded on solid relationships, became more important than ever for many organisations. Yet in some cases, viable alternatives for goods and services that were no longer available or affordable from the existing supply chain had to be sought. 

Supply chain and procurement models got reassessed. Some found the rapid changes needed to these models weren't supported by inflexible legacy technology solutions. A common approach to supplier management is to upload structured data into ERP systems during supplier onboarding. This data may be available across the organisation but usually gets siloed by department or geography. It also tends to remain somewhat static with occasional updates to product specifications or pricing data. This approach was never designed to look at options for viable substitutes in a rapidly changing world.

Instead, it ensured uniformity by restricting buyers to the 'preferred' supplier by default. In recent times this approach has proven to be limiting. If your existing supply chain is unable to meet demand or has applied significant price increases, the act of finding alternative suppliers lacks a data source.

Demand for ESG

There's another paradigm shift that is testing traditional supplier selection, the uplift in environmental, social, and corporate governance (ESG). Emerging ESG criteria and regulatory obligations are changing who organisations consider optimal suppliers. New ways of determining whether suppliers are a fit for the ESG imperatives and vision of an organisation are finding their way into many procurement processes. IT systems support is often lagging in this task.  

Most organisations collect a plethora of unstructured supplier information during a procurement process, such as responses and proposals to EOIs and tenders. This intelligence often lies untouched after supplier selection. Another rich data source often lies in emails between an organisation and suppliers as they communicate about doing business together. 

In addition, online sources of information about supplier capabilities, pricing, and product or service specifications continue to grow. How can these fragmented and unstructured data sources find suitable alternatives when supply difficulties, inflation, or new ESG expectations see the existing supply chain falling short? Retroactively loading unstructured supplier data into an ERP system is not the easiest answer. It defies the principle of only capturing structured preferred supplier information.

Enter: Generative AI

An expanding group of procurement specialists see the opportunity to tap into unstructured data sources and the advent of new technologies, like generative artificial intelligence (AI), is making it easier to do so. However, there are challenges. Any approach must be able to seek out information from a larger subset of suppliers than the default available in legacy ERP systems. Alternative suppliers must be a fit for existing supply chain dynamics. They must meet expectations for quality, timeliness, and cost, plus new criteria for issues such as ESG must apply. The approach must find suppliers that fit within a flexible and adaptable set of criteria – and quickly.

You may have noticed the recent upsurge in communications about the power of generative AI. AI is the simulation of human intelligence processes by machines, especially computer systems. Generative AI not only simulates, but creates new content such as college essays, songs, and digital pieces of art. The latest computer-generated art - particularly music and paintings - are popping up regularly in news feeds. It's hard to tell whether all the excitement that has come with the launch of ChatGPT in November 2022 comes from human- or computer-generated sources. The day of generative AI is upon us, and new ways are emerging daily that will transform every aspect of our lives.

A valuable approach when applying generative AI is to create a generative adversarial network (GAN). A GAN uses two or more neural networks to create outputs and then applies an adversarial discriminative model to evaluate those outputs. Feedback loops are built between the neural networks allowing them to learn from each other. It is conceivable that GANs could change supplier procurement by using neural networks to search for alternative suppliers in fragmented and unstructured data sets. The discriminator model could act as an adversary to the neural networks, trying to identify the best supplier based on a range of criteria.

An exciting development already taking place in a handful of cases is the link between GANs and geographic information systems (GIS). GIS adds another spatial layer of fragmented and unstructured data in which to allow the GAN to make decisions dependent on location-based criteria. Weather and climate change, geopolitical change, demographics, disaster management, transport planning and traffic prediction, and local planning controls are some of the data sets GIS offers when identifying alternative suppliers.

Impact on real estate industry

Our team at Lendlease are considering what this all means for the real estate industry. Our suppliers are an integral part of our success, and our relationships with them are crucial. Our digital platform, Podium, harnesses over 60 years of real estate experience. It provides data and insights across the property supply chain. Generative design helps identify the best supply chain options for property development. Predictive insights enable better tenant and place experiences for existing property portfolios. Supplier data is a fundamental component of Podium, with the platform actively supporting and encouraging supplier collaboration and co-innovation. The supplier data is open and accessible. Our ecosystem of suppliers is as valuable as the platform itself.

The Podium for Development product, enabled by the Podium platform, takes advantage of many data sets provided by suppliers. It incorporates geography-based data including pedestrian modelling, and wind and solar shading to understand their impacts on development. The product utilises AI to "solve" optimal development outcomes by identifying the most suitable suppliers to transform the vision of a developer into a sustainable, vibrant place. It uses supplier data to generate forecasts for bill-of-materials and construction costs to optimise new developments.

Identifying the best suppliers using unstructured data sources is one example of the changes we may see as the power of using generative AI comes into play. They may not be in place to reduce the inflation we are currently experiencing but will help us get ready for the inevitable next wave.

Colin Dominish is the head of podium services at Lendlease Digital. He is a customer-first digital native with over thirty years of experience in bringing the best digital solutions and expertise from around the world and applying them to infrastructure projects.