A More Human Agriculture
Will technology lead to a “more human” agriculture?
ChatGPT, a therapist and entrepreneur walk into a coffee shop…
On New Year’s Day, I sat in a small coffee shop with a group of friends reflecting and chatting over brunch.
It wasn’t long before the conversation turned to generative AI, shifting the emotional tenor of our conversation in a palpable way. We discussed the implications of generative AI for our jobs, society and how we spend our time. Some of us were more optimistic, determined to make the most of an AI-ridden future. Others succumbed to dystopian fears, predicting an end to human connection.
Thinking about it, this debate isn’t just about AI – it’s a decades old debate on how humans can wield emerging technology to create a future we want to live in. Will technology replace us? Can we still influence its impact on us? How can we use our collective decision making and choose how technology will impact our lives and the planet?
“A greener, more equitable world won’t be an automatic outcome of the digital revolution. It’s our duty to shape its development even as its pace takes our breath away.”
Today in AgPunk, we’ll explore how technology might impact the role humans play in the agriculture sector, including what it means for labor exploitation, on-farm decision-making and farm business communication.
Can there be agriculture without humans?
For millennia, humans have been terraforming the Earth in order to grow food, fuel and fiber. Agriculture arose in human populations more than 10,000 years ago. And, while other animal species do change their environment to control their food supply (firehawks in Australia, for example), we are the species responsible for agriculture.
However, the number of humans involved in farming has declined significantly in the last century due to mechanization and increased agricultural productivity. In the United States, for example, we’ve witnessed a 74% reduction in farmers and a 51% reduction in hired farm workers since 1950. This trend is reflected across high-GDP countries, as highlighted by the team at Our World in Data: “As countries get richer, the average farm size tends to increase.”
Developing countries who have not yet benefited from large-scale agricultural mechanization have larger populations of farmers. There are 570 million farms in the world, with 84% of them considered “small” - on 5 acres (2 hectares) or less. These farms contribute 33% of the world’s food supply.
Let’s assume all countries strive for high levels of economic prosperity, which means that the population of farmers will continue to decline across the globe. But, will humans ever be fully eliminated from the agricultural process?
Connie Bowen at Farmhand Ventures - an emerging venture studio that’s building the future of work in agriculture - thinks not. “People will still be on farms in 50 years,” she told me when we spoke earlier this month. And I agree with her sentiment, not only from a technological perspective, but also from a philosophical one. The idea of humans being completely removed from the cultivation of food is a dystopian one for me. My hope is that technology will actually bring us closer to the environment - by allowing us more time to spend engaging with the world rather than performing routine work tasks.
From Connie’s perspective, while people will still be on farms, the number will be fewer than it is now – whether it’s due to economic trends, the pressures of climate change on the number of viable workdays or increased automation where the technology hasn’t yet penetrated (specialty crops, developing countries).
She highlighted how it’s not only automation and robotics that are needed to shift labor in farming, but also policy changes and new models of farm labor companies.
Lastly, Connie shared that labor shifts in farming will enable the existing workforce to upscale. She spoke both of changes to the conditions for today’s farmworkers, and also the number of “knowledge” jobs in agriculture.
“If we do it right, automation ensures that creative people can compete for the knowledge jobs in agriculture, which would ensure that we continue to iterate in a positive direction…. What we actually want to change more than anything else is greater open-mindedness and willingness to experiment in agriculture. That is ultimately what we should be pushing the whole system towards.”
Automation: the change that has been felt
A farmer sits in her warehouse meeting with suppliers, strategizing on-farm experiments and tending to the farm’s books, while a friendly assortment of humanoid droids buzz in and out, navigating today’s gamut of farm chores.
While I love the idea of farm-friendly Star Wars droids, this isn’t typical of on-farm automation today.
Today’s scenes are:
A farmer catches up on market news while the combine drives itself across the field during harvest.
Automatic sprayers and drip systems run on schedule in high tunnels and vertical farms.
Field data piped into farm management software generates charts showing analyses.
Hourly weather reports pop up in your Widgets.
Single-task harvesters picking tomatoes or apples at lower speeds than workers.
Drones scouting and spraying fields on their own, with human supervision.
Automation of farm labor in the last 70 years is responsible for shifting the number of farmers and farmworkers in developed countries. This shift is more prevalent in commodity row crops than specialty crops, and in developed countries than in developing countries.
How might we learn from the way agricultural automation has shaped developed countries to mitigate any negative repercussions of this shift for developing countries?
For example, while the United States has excelled at agricultural production efficiency, we’ve done so at great cost to the natural resource assets (soil and water) that power production. Can developing countries better equip themselves during the mechanization of their food supply to protect natural resource assets by incorporating regenerative agricultural practices?
We can also contemplate the impact of automation on human exploitation. Perhaps, instead of orienting around how automation can yield the greatest efficiency and profit outcomes, we can apply the objective of reducing human exploitation and suffering through automation – especially in industries that are essential, like farming, where exploitation is overlooked for the sake of production.
One hard truth of farming is that it’s a difficult job that poses dangers to human health. The tasks are repetitive and working conditions are variable. Farmers and farmworkers regularly run the risk of having occupational injuries and heat stroke.
Even with this incredible upside of reducing exploitation, we must recognize that automation will also reduce the number of paying jobs in certain areas. It will, and it already has done so. In fact, even beyond agriculture, “existing technologies already allow automation of half of all activities people are currently paid to do.”
The job loss impact of automation is significant, with studies showing that every robot added per 1,000 workers in the US led to a 0.4% decline in wages and a reduction in employment by 0.2% relative to the total population.
“Automation technologies generally don't bring shared prosperity by themselves… They need to be combined with other technological changes that create jobs.”
An example of this can be seen in the retail sector. Reports have shown that automation (and AI) related job losses in the retail sector have been counterbalanced by an almost equal gain in downstream delivery jobs.
Better Conditions for Farmworkers: the change that relies on us
In the US, Farm Labor Contractors (FLCs) employ and manage farmworkers in a system that unfortunately exploits the poor and marginalized. While I recognize the benefits I get from this system – fresh, affordable food at the grocery store – I also lament that our farmworkers lack rigorous protections from exploitation.
This is because farmworkers tend to lack full citizenship protections. For example, the US Department of Labor’s National Agricultural Workers Survey (NAWS) in 2020 showed:
85% of farmworkers entered the US more than 10 years ago, yet more than 1 million of these farmworkers remain undocumented.
29% of farmworkers do not speak English.
Farmworkers average $13.59 per hour in wages with an average family income of $25,000 to $29,999. This is just barely above the poverty level for a family of four of $13.34, or $27,750 per year.
48% of farmworkers report having health insurance.
Separately, the Human Rights Watch found that sexual violence impacts women farmworkers, who have little protections in place at work to make the violations committed against them known and corrected. For example, while FLCs and those who employ farmworkers are held to OSHA employer standards, these standards are not regularly enforced.
So we exist in a time where farmworkers are essential but they are not adequately protected, and automation (which could shift the burden of farm labor from humans to machines) is not fully realized in all crop and livestock production systems. We must take action to rework our social and governance systems to look out for the workers that are braving harsh conditions to help us eat.
This is why we need improvements to farmworker wages, safe working conditions and access to job training resources.
Artificial intelligence: the change that is promised?
As she enters the equipment shed, the building whirs to life with a steady hum of intelligence. Her very own “Farm Jarvis” greets her with, “Good morning” and a report on today’s weather and chores. She hears Farm Jarvis’ assessment of their progression towards harvest, using the synthesis he provides to make a crucial decision about how she’ll approach marketing her crops this week.
A great example of supportive AI is that of Tony Stark’s relationship to Jarvis in Marvel’s Iron Man. In this relationship, Jarvis functions as a virtual assistant and data expert for Tony. Tony leverages Jarvis’ assessments to empower his own decision making. Tony can override Jarvis, and Tony can consult Jarvis when he needs a second opinion. Jarvis, unlike Ultron (a sentient, independent AI that ultimately decides to kill off humans), is in partnership with and ultimately controlled by Tony.
This is the future I hope we drive towards, one in which “AI should augment human intelligence, not replace it.” In this future, AI will be responsible for changing the ways that human farmers and farmworkers make farming decisions and interface with non-farming stakeholders.
For example, it’s been shown that yield projections are more accurate when crop simulation models are paired with machine learning models (a form of AI). As discussed in our first AgPunk article, simulation and model-based yield predictions are not nearly accurate enough to inform farmers’ decisions so they rely on historical performance instead. With AI, we could get to a level of computing power and intelligence where we can predict yields even more accurately than historical performance. This would provide farmers with more power and certainty around expenses (knowing exactly the best times to apply inputs, for example) and income (better marketing decisions and price decisions).
Additionally, while farmer poets do exist, many farmers do not enjoy writing (after all, they’re farmers). Yet, writing a business plan, a grant proposal or a website narrative are activities that support a farm’s business objective by helping secure capital, tell their story to stakeholders and align the farm staff around a common plan. What if generative AI such as ChatGPT could help non-writer farmers tell their story more effectively?
The power to choose
In his book Sapiens, Yuval Noah Harari describes our species as insecure. We occupied the middle of the food chain for most of our existence, and it was only recently that we’ve secured a spot at the top. We constantly search for threats to our survival. Unlike the confident lioness who’s held her place at the top of the food chain, we’re new here. We’re imposters.
And it’s right for us to be critical of the role technology might play in our future. As we create new technologies to improve the human condition, we have the power to design for more than just efficiency and profit. We can design technologies that strengthen our decision-making powers, reduce or even eliminate exploitation and maybe even give us more time to connect with the environment and one another.
While there isn’t a global governing body that stewards humankind’s technology roadmap, there are leverage points for shaping the way emerging technologies impact us. For agriculture, the influential players include farmers, policy-makers, thought leaders and technology companies. These players have the mandate to thoughtfully consider the future and ask critical questions:
Where are we headed?
Where do we want to go as a sector, and how might we get there using these technologies?
How might we anticipate and mitigate the negative repercussions, and better manifest our true positive potential through their use?
My hope for agriculture is that we are able to embrace technologies that help farmers run their businesses in healthy, sustainable and regenerative ways.
What do you think?
How do you hope that autonomy and artificial intelligence shape our farming future? How might we work even without these technologies to reduce the exploitation of human and environmental resources that result from agricultural production?
Contributors
Writing: Sami Tellatin
Contributions and Edits: Connie Bowen and Kristina Tober
Do you want to contribute to a future AgPunk? Let us know in response to this post. We’re open to publishing related fiction, or narrative non-fiction about technology, agriculture and sustainability.
Art and music update
Art: I’ve used AI- generated art for these first three installations of AgPunk. As I continue using it, I am researching more about its impact on human artists. If you have any resources or opinions on this, please reach out!
Music: Punk music has been one of my favorite genres, and the counter cultural themes of punk songs were part of the inspiration for “AgPunk.” As such, I’ve created a playlist for the newsletter called “AgPunk” on Spotify. I aim to fill it with upbeat pop-punk (with *mostly* positive messaging - hard to find punk that’s completely hopeful)! Please listen and let me know if you have songs you’d like to recommend!
Sami, I have some thoughts about technology in agriculture. This comes from 40+ years of participating in the food and ag space.
I continue to be underwhelmed by how short a distance we've come in the ag tech space. Agriculture is unique because, unlike most other production systems, it takes place outdoors. Vertical/indoor farming notwithstanding, the majority of variables that need to be dealt with in agriculture are in the realm of Natural Systems. Yet virtually every ag tech "solution" is really just a point solution, not a system's-level concept. And virtually all of these solutions are designed to work against, not with, natural systems. Battles may be won, but the war is already lost.
Unless and until "technologists" in agriculture develop a systems-level view, we will never get beyond
the First Order of point solutions. In logic systems, First Order logic incorporates "for all elements" into the syntax (i.e. point solutions) whereas Second Order logic incorporates "for all properties" (i.e. moving towards a systemic view). Second Order precedes Higher Order, and can include Set Theory. But that's getting ahead of ourselves. Right now we are stuck in the First Order box. Maybe we need the imagination of mathematicians to show us that there is much, much more to the world than what is being thought about today in agriculture?
This future is here! 💪
"For example, it’s been shown that yield projections are more accurate when crop simulation models are paired with machine learning models (a form of AI). As discussed in our first AgPunk article, simulation and model-based yield predictions are not nearly accurate enough to inform farmers’ decisions so they rely on historical performance instead. With AI, we could get to a level of computing power and intelligence where we can predict yields even more accurately than historical performance. This would provide farmers with more power and certainty around expenses (knowing exactly the best times to apply inputs, for example) and income (better marketing decisions and price decisions)"
Good read, nice work!