Part 3/5 of David Carlin's series on sustainability teams' use of AI today
Photo credit: Getty Images / Unsplash+
In his digest series Ask David, David Carlin explores the challenges sustainability teams face today, giving actionable advice on demonstrating financial value, strengthening business strategy, managing risk, and driving real organizational impact. This new five-part series offers insights into how AI can help bridge the gap between sustainability teams' broad mandates and limited resources. Click to read Part 1 and Part 2. In Part 3, David Carlin reflects critically on misplaced use of AI.
I am very sympathetic to sustainability teams here. There is enormous pressure across organizations to deploy AI, save money and, in some cases, reduce headcount.
Sustainability functions have already been under considerable pressure over the past few years, so that pressure can feel particularly acute.
But moving too quickly can be counterproductive.
If you start replacing human thinking or entire end-to-end processes with AI, you can very quickly lose control of the quality of the output.
We have been brought into situations where organizations have already tried to use AI for sustainability work and ended up with something generic and ultimately not very useful. In one reporting example, considerable time had been spent producing outputs that simply were not sufficiently tailored to the organization. Errors had also been introduced. The team then had to go back through everything with a fine-tooth comb to determine what was reliable.
Rather than saving time, AI had created additional work.
The broader evidence reinforces the need for controls. Stanford’s 2026 AI Index recorded 362 documented AI incidents in 2025, up from 233 in 2024, while noting that responsible-AI measurement is struggling to keep pace with rapidly advancing capabilities.
That is why starting with a well-defined use case is so important. What specific problem are you solving? Which parts of the process can AI genuinely improve? Where does human judgment remain essential? And how will you verify the outputs?
Opening up the box too early, automating a process before defining what good looks like, or placing too much trust in the output can quickly undermine the benefits. AI can be extraordinarily powerful, but without the right structure it can also allow you to produce mediocre work much faster.
For more information, visit the D. A. Carlin & Company or explore David Carlin's Substack.
Published by
investESG
investESG
