AI in Strategic Foresight 2025

Challenges to integrating AI into foresight practices FIGURE 5 0%10%20%30%60% 40%50%Which of the following challenges do you face in integrating AI into foresight? Lack of technical expertiseEthical or governance concernsResistance from leadership or stakeholdersDifficulty aligning AI outputs with human-centred foresightLimited access to relevant dataOther35%55% 44% 12% 8%39% 39% 27% 20%37%39%51% Those who have used AI Those who have not used AI Source: OECD Given these concerns, despite efficiencies, users have significant reservations about AI’s trustworthiness and ability to deliver truly novel or nuanced insights without extensive human oversight. These can be summarized by frequency of mentions as follows: 1. Output quality and trustworthiness (30%): The main concern is that AI often produces unreliable, shallow or unoriginal outputs. Respondents doubt its ability to deliver deep analysis, novel insights or credible foresight without human oversight. 2. Transparency and verification (12%): Practitioners struggle with AI’s lack of transparency and unclear reasoning. Because the sources and logic behind its outputs are hidden, significant human effort is required for validation and fact-checking, turning the process into an audit rather than an exercise. 3. Bias and data limitations (10%): Respondents worry about biased and narrow outputs due to AI’s dependence on English-language, Western and historical data. Its limited access to private or emerging information restricts global and forward-looking analysis. 4. Organizational and ethical issues (10%): Concerns extend to data security, unclear rules on ownership and citation, and the risk of over- reliance on AI. Many organizations lack formal ethical frameworks to guide responsible use. This is problematic – only 27% of experts using AI already said their organizations had formal ethical guidelines in place (Figure 6). 5. Methodological and usage challenges (10%): Using AI effectively is seen as difficult. Crafting strong prompts and maintaining quality requires expertise, with AI often reverting to simple or inconsistent outputs. The challenges outlined above indicate that AI is currently viewed as a tool that augments, rather than replaces, human expertise, requiring careful and continuous oversight to ensure the quality and ethical soundness of foresight work. AI in Strategic Foresight: Reshaping Anticipatory Governance 13