Some time ago, I wrote that AI is not a choice sustainability functions can put off, and that efficiency is not where its greatest value lies. A survey by Watershed shows that 43 per cent of respondents use AI, but mainly for tasks close to reporting, such as drafting reports. This highlights a gap between how AI is used today and its potential in more consequential areas of sustainability work. The question is therefore not whether AI has a role to play, but what it will take for us to take the next step.
In its report State of corporate sustainability 2026 – AI edition, Watershed, which provides a platform for climate and sustainability work, surveyed more than 230 sustainability professionals about their use of AI, among other topics.
The report shows that almost half use AI, but mainly for tasks that are close to the existing reporting process and relatively easy to define and control. That is not surprising in itself. It is natural to start where the risks are lower and the benefits easier to see. At the same time, however, the figures highlight the gap between potential and actual use. AI has entered sustainability work, but it has yet to play a significant role when data is turned into priorities, supplier engagement and tangible emissions reductions.
Here are some insights from the survey:
The next step is about trust
When AI is used in more consequential areas of sustainability work, trust becomes central. Among those who do not yet use AI, some of the most common barriers are:
The interest is already there. One of the clearest barriers is a lack of trust in the data, the output and the way of working. The next phase therefore needs to be less about showing how quickly AI can work and more about showing how the output was produced, what has been checked and where human responsibility still lies.
Trust needs to be built into day-to-day work. It needs to be clear who reviews AI-generated output, how uncertainty is handled and who is accountable when that output is used to inform a decision. Trust therefore also becomes a question of how we organise the work and allocate responsibility.
From time savings to business value
AI can free up time otherwise spent on data collection, administration and reporting. These tasks currently take up a considerable amount of teams’ time. At the same time, the picture is more nuanced, as only 21 per cent of AI users report significant time savings. Introducing AI into an existing process does not, in itself, guarantee a greater impact. To realise its potential, technology, data, accountability and ways of working need to fit together.
When technology needs to fit into the bigger picture
For small sustainability teams under pressure, the question of AI goes beyond the choice of tool. It is also about how technology, data, skills and ways of working need to develop together.
The World Economic Forum’s Chief Sustainability Officers’ Outlook, published in September 2026, shows that skills are an important part of the wider picture. Of the 103 chief sustainability officers from large companies who took part, 68 per cent expect gaps in skills and capabilities to limit their organisations’ ability to use AI to advance sustainability over the coming year.
The report’s analysis emphasises that realising the benefits also depends on reliable data, clear governance and collaboration across sustainability, technology, finance, operations and risk management. AI is described as a means of supporting the delivery of the sustainability strategy, rather than a substitute for it.
For sustainability teams, this raises a central question: how do we build ways of working in which AI-generated output can be reviewed, understood and used?
McKinsey’s global AI survey from August 2026 also points to a gap between individual benefits and organisational impact. Eight in ten respondents say that AI has improved their own productivity, while 37 per cent report a positive impact on their organisation’s operating profit. Among the organisations McKinsey identifies as the most successful in using AI, almost three quarters have fundamentally redesigned their workflows, compared with around a quarter of the other respondents.
The survey spans several industries and shows correlations rather than definitive causal relationships. Even so, it reinforces the argument that AI needs to be accompanied by changes in ways of working if the benefits are to reach the wider organisation.
This means that automation needs to be combined with clear choices about quality, accountability and use. The ability to bring together technology, measurement, ways of working and quality assurance will determine whether AI remains a tool for more efficient reporting or can also support more consequential decisions.
Sources