The future of agriculture as seen by 300 synthetic personas
With AI lowering the barrier to building, the question that remains is what’s the right thing to build. Product development is messy. Signals come from all directions; existing user data, audience research, emerging policy and market analysis. Taking a longer view, strategic foresight takes a structured, systematic, and participatory approach to exploring, anticipating, and shaping the medium to long term future to inform current decision-making.
Focus across multiple horizons
The key difference between a product roadmap and strategic foresight is that the former is execution focused, while the latter is focused on identifying and assessing different “horizons” or stages of future opportunities.
But the challenge fast moving teams and businesses often face is having the space and time to focus on multiple horizons. AI also promises to reduce the resources required for this type of work, creating the opportunity for strategic foresight methods to be integrated into the approach of any org size.
‘Synthetic foresight’ uses generative AI and agentic systems to accelerate, deepen, and automate the process of exploring potential futures.
Personally – I was curious but skeptical as to the value of such an approach. So I adapted a project methodology to run a research project and generate a report into the future of agriculture in an age of ubiquitous AI.
The project involved generation of 300 unique personas, each with their own background, points of view, and conversational traits, 20 ‘digital twin’ experts based on real people and 3 simulated 1 hour workshop discussions.
Just More AI slop?
The outputs and scenarios generated are certainly thought provoking, alarming even and do bear resemblance to the types of reports and thematics seen across the industry. But it's hard to understand whether this is just a result of referencing and referencing content found online rather than based on lived experience.
In this sense, is recycling existing information with a veneer of McKinsey speak offering anything new? Is this just more AI slop? Possibly it is. But this limitation points towards where the opportunity lies to unlock more value from an AI mediated approach to strategic foresight.
Opportunities
Integrating real voices and real people to this approach (ie composite audience personas) is where the real value could lie.
Aggregating user interviews, workshop outputs and existing domain knowledge into accessible systems provides a much more solid and representative foundation for insights.
Beyond this, the approach doesn't limit us to human voices and input. Publicly available datasets can be used to provide other non-human perspectives. Future 'workshops' could include more diverse ranges of participants - human and non-human.
The project
This experiment and the report is both an output and an investigation into the tools emerging technologies available.
The methodology is based upon a forecasting project on the future of journalism from the Tinius Trust. Prompts and methodology are adapted from this project, and applied to the context of agriculture in an age of AI.
The experiment looks to explore some of the plausible future scenarios for agriculture, and also to examine the capabilities and opportunities that new tools present.
The entire report was generated through a combination of Google Gemini and NotebookLM. Sections in black are notes, written by a human (with the assistance of speech-to-text transcription).
See the report here