AI has arguably passed the Turing Test: AI engines such as ChatGPT give answers that could very well be a human’s: Albeit it still needs – no doubt – a good read and critical review of answers to avoid accuracy and correctness pitfalls. Sometimes indeed even the insights of a subject matter expert to ‘get it right’ – just as would be for any human conversation partner.
And yet, we’re just in the early stages of finding out where AI, in its most advanced forms, is going to be really, really good at. Not ‘hyped’ good, but useful, sensible and with genuine benefits – which may (of course?!) be of good or bad nature.
No insight could summarise the current status quo any better than the following statement on X (formerly Twitter) by Joanna Maciejewska:
The statement is entirely to the point.
The good news is, that at least for now, in the field of sustainability, AI excels in some areas that are fundamental to the discipline, to creating positive impact and progress … yet where we human beings tend to get frustrated with, and loose a lot of our valuable life time: Chasing and compiling data, writing reports, cross checking data etc.
To push the envelope a little bit, and to put it into a different set of words, too: at this very moment, AI applications can be good at supporting companies in particular at ‘doing less harm’ …. but less so when it comes to ‘going above and beyond’ and hence creating genuinely positive impact.
AI complements Big Data rather well
One of such reasons is because AI complements Big Data rather quite well, as it adds on the ability to ‘work with’ Big Data much more efficiently than most applications so far were able to do. For example, AI is great for gaining insights from unstructured data – of which there is a not only a lot of in the Sustainability field, but which is also the reason why many of us professionals spend hours, days, sometimes months, crawling through data sets in order to find the relevant insights., and to figure out how to structure them, and turn them into impactful actions for change.
Hence, this upcoming series of blog posts is a first stab at figuring out where AI is already potentially beneficial to us Sustainability Professionals, and the companies we work for. As AI matures, I am in no doubt though that further, and overall much more interesting, application areas can and will be found.
But for now, there is no reason why any AI investment in a company should not also consider the sustainability field as potential target application. To that reason, we’ll be looking into three particular areas:
- Sustainability Reporting, not the least with a view to the first CSRD reports being worked on and published in brief.
- Supply Chain Management, with a view to compliance and audit validation in particular.
- Financial Reporting, with a view to how AI can make it easier to financially quantify sustainability-related measurement, and help in this way to align and integrate the two reporting worlds, under the long-shot assumption of them ultimately becoming ‘the same’.
AI’s raison d’etre: Turning data into value
But before any of that, let me discuss in brief on the different types of AI applications possible, or useful, in the corporate context, and also where most of the cash and effort goes (if at all) in the current moment).
First off: what is the fundamental intent behind AI?
I am pretty sure most people will come up with a vast array of interesting ideas. However, fundamentally the intent is rather quite simple: ‘Turning data into value’.
Whatever we know and have experienced in regards to AI, comes down to the fact that somewhere, someone, saw a value in turning existing data into something ‘else’.
Where it gets interesting is the interpretation of ‘else’. Because thanks to the processing power we have available in the current time and age, this ‘else’ starts to take on creativity-like shapes, as explain in the below illustration.
All of that has been part and parcel of research in the past (cf. Turing Test mentioned above, or the ‘Put that There Project’ from MIT), but it is only recently and thanks to processing power available to us at the present time, that we are reasonably successful at coming up with use cases that are of interest to any Jane and Joe Doe – and that hence triggers them to start using and benefitting from this technological option.
So far so good. Let’s go back for a moment to ‘Turning Data into Value’.
What shape and form, or what trajectory, does such novel type of value generation typically take? The following chart, giving us some insights into the tech start up world of Unit8 from Switzerland, and their (very much tech forward) customers’ demands.
Applying the above to the Sustainability field as well as to the present time, combined with the currently most pressing challenges we have at hand, Step 2 and Step 3 are where nearly every company in the world could benefit in their Sustainability efforts from the application of AI.
- To make the insight snappy: finding, organising, aligning, and understanding data – as well as, possibly more importantly, figuring out where data gaps still exist after unstructured data ‘out about’ is taken into account.
- Making it snappier still: Asking for, evaluating, benchmarking and analysing Compliance -related data – of all types, in just about any structured or unstructured format we can think of.
Of course there will more, better and more interesting applications coming up in the future. But that’s not the point. We can start here and now, pretty much immediately.
It’s not just that: for most initial use cases in the Sustainability space, ‘off the rack’ vendor AI software packages do exist. Only – it’s not the same vendors or tools you already use for your SCM, PLM, etc. Because those vendors are mostly behind the curve on the topic.
And that’s probably the most important point to keep in mind:
- Previously, we had SCM / PLM … subject matter experts building excellent, but ‘clunky’ tools. Very specifically targetting a clearly structured process. Complexity arose where such processes varied between companies, products, suppliers.
- Now, we’ll start to see tools that can deal with all of those topical areas, built by people who do not understand process in depth, but how to make the best of any data you may have from product, supply chain etc. Complexity arises in this case because the data is not always clearly associated with the why and the where it has been generated.
The magic happens when we bring process / subject matter experts and data experts together. To figure out how to do more of, and better, what already exists.
Which means: AI is not a solution on its own.
It can be a great addition, augmentation, of the efforts we’ve already invested. And it may be able to tackle some of the challenges we did not quite know how to approach, because the overhead to solve them was just not financially (and otherwise) viable.
Like any novelty: it’s not going to be cheap per se – but viable for the first time at all.
And for a whole range of use cases, that precisely is the value add. The big step to start with.
How this could concretely look like in the 3 areas mentioned above? We’ll bring that to you over the weeks and months to come.
