GOOGLE CLOUD NEXT, LAS VEGAS – If you could use technology to grow greater volumes of more calorie-dense food on the same amount of land, would you? That’s exactly what Ohalo Genetics is trying to do: use artificial intelligence (AI) to analyze and alter the DNA of plants to maximize plant yields.
As an aspiring backyard farmer, I may be biased, but this was one of the coolest things I heard in the sea of news that came out of Google Cloud Next.
If you’re familiar with DNA sequencing at all, you know it basically consists of a series of letters that can arranged in different combinations to yield different characteristics. The unique sequence is known as the genotype. The physical characteristics that come from those combinations are known as the phenotype.
During a press panel, Ohalo’s founder and CEO David Friedberg explained the “key challenge in plant breeding is [figuring out] what genotype, what sequence of letters predicts what phenotype.” Ohalo’s goal, he continued, is to use generative AI to discover what genotypes map to each phenotype.
Once they know that, they can use DNA editing tools like CRISPR to alter the plants to produce desired traits. These could include more calorie-dense vegetables, plants with higher yields or resistance to certain diseases and climate or soil conditions.
“That will be a massive boon for human societies, for the planet, for sustainability and so forth,” he said.
Friedberg noted Google has been a pioneer in this field, pointing to its decision to open-source its DeepNull model – which is focused on phenotypic prediction – in 2022.
DeepNull gave Ohalo a fair idea of what it could do with Google systems and infrastructure to realize similar outcomes at a larger scale. Specifically, he said Ohalo is working to process sequencing data for hundreds of thousands of plants as well as phenotype data from images of plants in both visible light and other light spectra to better understand how the two are related.
“I think if you fast-forward a couple of years, we could find ourselves in a situation where we say things like ‘I would love to have a corn plant that is twice as big, grows in this part of the world, has kernels that are this big, is resistant to the five diseases that I specify and can grow in four inches of rain’ and the model can predict the exact genomic sequence that can realize that plant and then we can actually execute and deliver that plant to the environment,” he concluded.
This to me is a really cool application of AI tech, and one which could help us feed the planet’s ever-growing population.
Hurdles on the AI horizon?
That said, I’ll be interested to see how it overcomes a few potential hurdles. In the U.S., the Food and Drug Administration (FDA) has a long history of approving genetically modified organisms (GMOs) for sale in grocery stories. It even recently issued fresh guidance for companies looking to bring genome-edited plants to market.
But still, building the plant is one thing. Getting regulatory approval to sell and profit off said plant is another.
The second issue I can see is around intellectual property. Will the AI-created foods be patented in the way that major food producers (ahem, Monsanto) has with other GMO products? Maybe. And that could create more headaches for farmers.
Finally, there’s the potential for consumer backlash. Folks are already plenty wary of GMOs and when you throw big, spooky AI into the mix, there’s no telling what the reaction will be. Marketers will have their work cut out for them.
Still, this use case speaks to me louder than the ones I heard about for ad agencies or big banks. Maybe because the fastest way to a person’s heart is through their stomach.
Catch all of our coverage from Google Cloud Next 2024 right here.