Recently, Janek Bartel, our crop science laboratory manager in Canada, shared his insights on the continued importance of human expertise in seed analysis in an issue of Seed World Canada.
Janek’s team is working with multispectral imaging, which enables a more comprehensive examination than the human eye can provide and can reveal critical information that might be missed by traditional methods. Janek acknowledges that while advanced technology, such as multispectral imaging and AI, allows us to go even further for our customers, “we still need talented, knowledgeable human experts in the loop – and that will not change anytime soon – if ever.”
The data that imaging tools produce is often combined with algorithms and, as Janek says: “These models are limited by the data they are trained on. We are far from the point where they can figure things out on their own, especially in niche fields like seed analysis.”
“Right now, we are in the wild West of AI. The idea of AI spontaneously developing the intelligence to sort seeds autonomously is not here yet, because we need high quality data and someone to train the model.”
Recently, Janek spent 30 hours identifying features and classifying a few hundred images in order to train a demo model. A more accurate prediction could require the classification of tens of thousands of images, and the model still might struggle to identify simple differences between seeds, which would call for a reevaluation of factors such as algorithm complexity and the quality of data collected.
“That is where human expertise comes into play. No matter how well the model performs, it lacks one crucial thing: common sense. It is simply evaluating patterns and data against the dataset it has been given as a training set.”
Humans, with knowledge of environmental factors and real-world implications are needed at the controls, to see beyond the basic data. An experienced analyst is needed to provide context that machines cannot.
Janek states: “We are in a phase where AI and human analysts need to work hand-in-hand. The machine learning models are powerful, but they are not all-knowing. And while it might be tempting to think we can automate everything, human judgment is still irreplaceable.”
Read the full article as previously posted in Seed World Canada.
For further information, please contact:
Janek Bartel
Crop Science Laboratory Manager
t: +1 800 952 5407
About SGS
SGS is the world’s leading Testing, Inspection and Certification company. We operate a network of over 2,700 laboratories and business facilities across 119 countries, supported by a team of 99,250 dedicated professionals. With over 145 years of service excellence, we combine the precision and accuracy that define Swiss companies to help organizations achieve the highest standards of quality, safety and compliance.
Our brand promise – when you need to be sure – underscores our commitment to trust, integrity and sustainability, enabling businesses to thrive with confidence. We proudly deliver our expert services through the SGS name and trusted specialized brands, including Brightsight, Bluesign, Maine Pointe and Nutrasource.
SGS is publicly traded on the SIX Swiss Exchange under the ticker symbol SGSN (ISIN CH0002497458, Reuters SGSN.S, Bloomberg SGSN:SW).
