Artificial intelligence (AI) is everywhere these days, and plant breeding is no exception. As machine learning and genomic prediction take over the industry, there are several things you can do to ensure that you will succeed in a field experiencing unprecedented change.
1. Learn the language of data: Do you need to learn how to write machine learning code? No, but you do need to understand what genomic prediction models can and can’t do. What does heritability mean in the context of a model? What does it mean to validate a prediction? Why does a model trained in one environment stumble when generalized to another? Knowing the answers to these questions is the difference between thriving and being left behind.
2. Treat your data with the same care as your germplasm: Algorithms can’t compensate for decades of messy phenotypic data, deficient traceability and/or inconsistent storage formats. For your breeding program to succeed, you need to standardize how the environment of each trial is documented, how phenotypes are recorded and how they connect to genotype. No one will claim it’s glamorous work, but these are the practices that will reward you with strong genomic predictions.
3. Stop thinking in silos: And start thinking collaboratively. If you’re a breeder, make friends with a bioinformatician. If you’re a plant geneticist, get out in the fields with an agronomist. The future belongs to those who step outside their own expertise to embrace systems thinking and cross-functional collaboration.
4. Get comfortable with calculated uncertainty: A prediction model won’t give you absolute certainty, but it will give you better probabilities than what you were working with before. If you insist on perfection before making a decision, you’ll fall behind the competition with blinding speed.
5. Keep an eye on industry trends: What platforms and teams are leading the technology revolution? How are they building their genomic prediction and automation capabilities? When you understand how the industry is evolving, you can anticipate where your own role is heading.
6. Be the person who pushes for change: Some people adopt new tools only when they have no alternative. Others demand them before it’s obvious that they’re needed. Don’t wait to learn genomic prediction or machine learning. Ask to join a pilot. Volunteer for a project you don’t totally understand. Define the future before someone defines it for you.
