Three years after the launch of ChatGPT, businesses are increasingly recognising that artificial intelligence brings both major opportunities and serious environmental, social and governance (ESG) risks. AI can improve areas such as climate risk analysis, supply chain transparency and access to information, but it can also contribute to rising data centre emissions, deepen inequality and create unintended social harms. The article describes AI as a “jagged frontier,” meaning it can be highly effective in some areas while unreliable or harmful in others, especially when applied to sustainability challenges.
Dr Louise Drake of the Cambridge Institute for Sustainability Leadership argues that AI adoption in business is now unavoidable, with many executives already using generative AI regularly. However, she stresses that AI should be used to enhance human thinking rather than replace human judgement. According to her, judgement is essential because it involves taking responsibility for decisions, and handing that over to opaque systems can weaken accountability, ownership and critical oversight.
A major concern highlighted in the discussion is that AI-driven sustainability decisions can become flawed when the data used lacks context. Drake explains that context provides the real-world understanding needed to interpret environmental, social and economic interdependencies. Without it, business leaders risk making “systems-foolish” decisions that ignore how actions in one area can negatively affect others. She believes context helps leaders become more “systems-intelligent,” enabling better long-term and more grounded decision-making.
Drake also reflects on how she and her colleagues use AI in teaching, research and writing. They test its capabilities in tasks such as finding literature, analysing data and generating summaries, while constantly comparing outputs against their own expertise. Although AI can improve efficiency, she warns that overreliance on it can weaken the quality of insight. She describes a “conceptual smoothie” effect, where AI-generated content may appear polished but loses nuance, depth and the relationships between ideas.
In education, these risks are especially visible when students rely too heavily on AI-generated outputs, including fabricated references and shallow analysis. Drake notes that such work is not only academically penalised but also demonstrates poorer critical thinking. She believes the same dangers apply in professional settings, where business leaders may be tempted to accept plausible-looking AI outputs without properly interrogating them. This makes domain expertise and critical evaluation more important than ever.
The article also explores how AI is reshaping sustainability roles within organisations. As companies increasingly use AI for automatable tasks such as sustainability reporting, there are concerns about reduced opportunities for junior- and mid-level sustainability professionals. Drake suggests that organisations are still experimenting with AI’s role, but role redesign and career pathways have not yet caught up. She sees this as both a risk and an opportunity, as sustainability professionals may be able to move away from repetitive reporting tasks and toward more strategic, enterprise-level work.
Rather than focusing primarily on reporting, sustainability leaders may need to develop stronger skills in AI literacy, systems intelligence, futures thinking and commercial understanding. Drake suggests that AI could help reposition sustainability teams as more analytical and strategic functions, where professionals spend less time producing data and more time interpreting it, linking it to business strategy and supporting long-term value creation. However, she also warns that organisations need to think carefully about how these new skills will be developed and where future talent will be trained.
Among the most common mistakes business leaders make when using AI for sustainability decisions is assuming that outputs can be trusted without scrutiny. Drake points to the classic “garbage in, garbage out” problem, noting that AI outputs are only as good as the inputs and training data behind them. She also warns that leaders often fail to define sustainability goals clearly enough, focusing on narrow sub-goals such as reducing deforestation or eliminating child labour without fully considering the wider systems in which those goals sit. This can result in AI simply optimising isolated targets rather than supporting broader sustainable transformation.
To counter this, Drake emphasises the need for strong critical thinking and interdisciplinary learning. She encourages students to question where knowledge comes from, how it is created, and whose perspectives may be missing. Since AI reflects the biases embedded in its training data, it can reproduce dominant or Western-centric viewpoints while overlooking marginalised experiences and local realities. For sustainability leaders, understanding these blind spots is essential to making more inclusive and effective decisions.
The article also raises concerns about AI being deployed without a clear sense of purpose. Drake notes that many organisations are adopting AI quickly out of fear of missing out or in pursuit of short-term efficiency, rather than asking what the technology should ultimately serve. Because AI is fundamentally an optimisation tool, it can reinforce existing unsustainable systems if used carelessly. Without clear goals, it may improve efficiency while worsening issues such as inequality, job loss and concentration of power.
Looking ahead, Drake believes the most important leadership capabilities in an AI-shaped world will be critical thinking, contextual awareness, systems insight and the ability to exercise wise judgement. She also highlights the importance of storytelling, relationship-building and influence, arguing that leadership is not only about technical analysis but also about shaping shared purpose and motivating people. Since AI can reinforce echo chambers and deepen social divisions, sustainability leaders will need to use it intentionally to strengthen collaboration rather than polarisation.
Overall, the article argues that AI should not be seen as a simple solution or threat, but as a powerful tool that must be used with care, clarity and human responsibility. While it can significantly improve efficiency and support better decision-making, its value depends on how thoughtfully it is deployed, what goals it serves, and whether leaders maintain the human qualities needed to guide it wisely.
