AI: removing jobs or enabling change?

Grappling with the challenge of complexity

The reason why this is relevant is that the Bega Group is a FMCG company within the food and beverage industry with many iconic brands loved by Australians. Procurement spend is over $2 billion annually, with 18 domestic manufacturing sites spread across an area the size of Europe. It’s a large and complex business.

Used correctly, AI should make us more efficient. In practice, we need to validate data and arrive at a reliable data set, because only then will AI be able to support us in making decisions and for the procurement team to deliver maximum value to the business. So while I’m happy to use it for these reasons, in practice, there are a lot of problems that require human intervention, as per the red bucket analogy above.

The doom and gloom about the effects of AI on entry-level jobs is a pessimistic view. A hundred years ago, an entry-level job was working in a knitting mill, sewing two things together. When I started my career, it was a time of embracing digital data entry and learning how to use email and in time, it’ll be different again.

The graduate I hired most recently is phenomenal. She’s 21, a digital native and uses tools like ChatGPT as naturally as breathing, while I know business leaders who struggle to use Teams, let alone AI. There are still going to be entry-level jobs. They’ll just be different jobs that demand different skills.

"There are still going to be entry-level jobs. They’ll just be different jobs that demand different skills"

The need for human input

It’s important to acknowledge that companies will still need people to buy their products. Most FMCG businesses are not only selling their products to Boomers they're selling it to every generation, including Millennials and the upcoming Gen Alpha. And if the younger generations are locked out of jobs taken up by AI, and therefore have limited money, the capitalist economic model is not going to work.

If that generation aren’t earning a living wage, they can't afford to buy houses, they can't afford to move out of home, to buy healthy food or participate in leisure activities and socialising. They'll have nothing to aspire to. Our governments will therefore need to plan for how they manage the impacts of AI. Developed economies are highly unlikely to accept a whole generation of individuals who have limited economic participation as a result of being born at the ‘wrong time’ due to AI evolution.

The need for humans working alongside AI is further underlined by the things that it can’t do, quite apart from reconciling data. A lot of the art for our procurement teams is being able to build, maintain and nurture relationships. It’s why our supplier’s partner with us. Because we can communicate the wider vision and journey of our company to them in terms of the impacts to people. As a result, they feel good about dealing with us. They might be able to get a better financial outcome from someone else, but we’ve provided value and connection beyond the financial. That’s not something a machine can do.

Here's a working example of this. One of our factories went down on a Saturday at three o’clock in the afternoon. To keep operations going, we had to get bottles freighted from one side of the country to the other. Our Head of Packaging used the strength of our relationship to resolve the issue. Not only did the supplier take the call on the weekend, but we had agreement bottles freighted from Queensland to South Australia in under 15 minutes.

That feels like it’s a situation that AI will never be able to replicate. Another concern is with intellectual property. We have some of the most loved and well-known brands in Australia. These are well-known names, not owned by global food giants, and it’s important to us that we have control over them.

For example, if we were working with a packaging agency and they provided a packaging design for Vegemite using AI, the AI company could claim that they own it. Given that this is one of our brands, we own the intellectual property, but if it isn’t human-generated, the ownership of the intellectual property can be challenged.  

This means that wherever AI is used, care must be taken and obligations around agreed usage confirmed. We need to be clear about where agencies can use AI to speed up the efficiency and effectiveness so we’re maximising AI efficiencies for development work, but ensuring the finished artwork is human-generated.

Making AI a sustainable solution

There are areas where guardrails remain ineffective. In my substantive role as our Ethical Sourcing and Sustainability Lead, the environmental impact of AI is not sufficiently understood.

There are parts of Texas that are in drought, and yet they are building data centres there that use a disproportionate amount of water just to cool the computers down. It means the conversation is partly about who has the capital to set up the data centres, and who has enough funds to secure water and power rights.

Much of that power comes from fossil fuels, but there are other implications. When you see a company say, “We are sustainable,” you need to question their AI usage and how many microprocessors they are buying. How many tracts of earth had to be cleared to get the rare earth minerals that are in them? And how much processing by-product is being put into the rivers?

Despite this, enormous positives in terms of efficiencies remain and we’re seeing those across the business. Our finance teams are utilising it daily to review data inputs, to identify areas of concern and suggest recommendations. Even I’m using it more frequently to transcribe meetings, provide meeting notes and action lists.

People can also see it reshaping the legal function. Maybe you could get AI to show you the difference between two contracts. Or you could use it in negotiations to extract more value from conversation and contracts. Those are both plausible scenarios.

I use AI all the time, and it’s clear that it can do some incredible things. For that reason, the technology is going nowhere. That isn’t a bad thing. But it has implications for all of those who choose to use it, whether they’re recruiting people to work in procurement, managing environmental impacts, intellectual property, or trying to understand how many red buckets they need to buy.