Artificial Intelligence (AI) is increasingly shaping the future of AgTech, but for small and medium-sized enterprises (SMEs), understanding how and when to adopt it remains a key challenge. Our first webinar, in collaboration with James Cogley, Principal Applied Science Manager at Microsoft, addressed this by demystifying AI and offering practical guidance for businesses navigating its adoption.
AI is software that learns from data to make predictions or decisions. It is only a tool, one that can produce flawed outcomes if trained on biased or incomplete data. The hierarchy within AI goes from broad AI systems to machine learning (ML), and further into deep learning, which uses complex neural networks. Foundational models, pre-trained on vast datasets, are a major recent development, enabling businesses to build applications more efficiently. Rules-based systems (as opposed to ML systems) are predictable and easy to audit but can become rigid and difficult to scale. ML systems are more flexible but often lack transparency, making their decisions harder to interpret, or trust.
James and our audience also explored the ML pipeline, emphasizing that successful AI implementation depends on more than just algorithms. It requires collecting and cleaning data, training models, validating them with test data, and continuously monitoring performance after deployment.
In AgTech, AI is already being applied in areas such as advanced analytics, anomaly detection in crops or environmental conditions, and computer vision for monitoring animal health. However, SMEs face several barriers, including limited access to high-quality data, concerns around explainability, connectivity challenges in rural areas, and a shortage of specialized talent.
A simple three-step framework can help deal with these issues:
- clearly define the business problem,
- assess whether sufficient quality data exists,
- determine whether AI is the right solution compared to simpler alternatives.
And a major piece of advice: do not adopt AI for its own sake!
Due diligence is crucial: data privacy, licensing, and the differences between enterprise-grade and consumer AI tools must be carefully considered before implementation.
With input from our members on other AI related topics of value to them and their businesses, we hope to develop from this one-off webinar to a series.
Watch this space!
