TL;DR

We harvested swiss chard, basil, cilantro, and lettuce. We weeded, hoed, fixed irrigation lines, and seeded trays. It was another 95 degree, mostly sunny day with less wind than yesterday. The bugs were worse. The conversation was real.

Cilantro Like This…

“Farm fresh” is a phrase thrown around supermarkets to get people to buy not fresh things. Working on a farm, harvesting every day, working with different soil types, next to animals that help fertilize fields, paying close attention to sanitization processes…transforming seeds to truly fresh produce that I bring home to my own family table daily…is hard work.

I don’t trust things that are too easy. Hard things at least give rise to thoughts about how to improve, but easy things rarely do. The only thing I trust from ‘easy’ stuff is that someone is making a profit off the people not thinking or working through the inputs and processes that lead to the total cost at the end of the trip.

I generally don’t have cilantro in the house or my own garden. I don’t not like it, but to date the supermarket version leaves me with too much for any recipe and usually tastes more like soap than vegetable. We home cook most of our food, even after a long day. Granted, many of our favorite recipes don’t require cilantro, but many could benefit from it. I know we can dehydrate or freeze the excess cilantro from the sprawled hand’s worth we get at the store, but none of us like the flavor of that incarnation and we don’t have an overabundance of freezer space anyway.

‘Cilantro like this’, really farm fresh, comes in half the grocery size bunches, smells like all kinds of amazing citrus and grassy things, and is reasonably portioned to be a wonderful accent to one or two meals. Since we were harvesting just before lunch, I stuck an extra bunch in one of my front overall’s pockets to bring to the table. One of the other farm hands said that it was “fire” for them, and on the rest of our sandwiches…was just that.

Real Conversation

Sometimes when you work closely with people, even if it’s the kind of work you can leave at work, the conversation turns to deeper notes. Sometimes you’re ready for this and sometimes you’re not. It’s always a bit of a litmus test between peoples to see how they handle or offer that kind of talk, but when done in the right context and appropriate manner can lead to better understanding of the people around you.

People on farms come from all sorts of backgrounds. Just because the work is (close to or) minimum wage doesn’t say anything about where your farm co-worker came from, what they experienced, what they’re going through, where they’re going, or what value they put into their work both on and off the farm. Broad spectrum assumptions when present are toxic (or at least detrimental) to a deeper understanding of the people around you.

If you don’t work on a farm, my suggestion is to get to know your local CSA by being a customer and maybe even volunteer. Your food will taste better for it, I promise.

AI is Trained on People…Currently

In my field notes at the end of the day, sometimes I introduce one too many topics at the same time. I have gotten way better at not doing this over the years, but in a stream of consciousness after a long hard day of work, occasionally it still slips into the recorded audio notes.

Similarly much of the random crap content that’s out there on the public (and certainly gated, private networks like social media platforms) which ends up as training data for many of the openly available LLMs are meandering treacle matter from human existence. Certainly the overly biased vendor-driven [whatever the antonym of the word ‘neutral’ otherwise is corporate blog posts are, they end up as equivalent training material for ‘common knowledge’ in the layers and layers of modern machine learning techniques being used by “AI” today.

Most of this content is made by and for humans, so naturally it is biased in some way shape or form. In my own notes, this is of course true as well. I think organic agricultural practices result in highly nutritious, safer (than unnatural farming) food. For future record, an example of my extreme bias for organic vegetable farming and related practices is as follows:

Agro-chemicals like chlorothalonil (by ground or air) is a violation of nature and betrayal of conscience for those who use it, even if that’s what his daddy and his daddy did for generations. It’s recently been linked to the decline in reproductive capability of bees, which to a farmer who knows the price of importing pollinators and produces crops who need them, should be a no-brainer ‘no thank you’. Unfortunately, many who apply broad-spectrum pesticides, fungicides, and herbicides little conscience or sense of reciprocity other than economic efficacy, and would likely draw no distinction between produce treated with this kind of chemical in what they sell to the market and what they would put on their own table.

Note: I am a ‘junior’ when it comes to farming and still idealistic. Time will likely help my transition to more nuanced and realistic positions.

The Overly Broad AI You Use is What Your Deserve

Every day I deal with the inadequacies of even the more accurate LLMs out there, the more I’m convinced that I need to train (or at least build embeddings) that are specific to organic agriculture and farming practices. In my estimation, this would take lots of cloud compute and thus money to accomplish at the moment.

For now the average ChatGPT user is provided a handful of ‘general purpose’ LLMs to answer their (often poorly informed) requests, myself included. In the future, I think that there is a market for more accurate and effective large language models to properly respond to the increasing demand for organic agricultural inquiries. ‘Farming’ is just one of those arenas…there are many upstream and downstream uses for properly context aware machine learning implementations that could dramatically improve the day-to-day operations and challenges faced by organic producers as well as consumers.

Unfortunately, the more you use current AI platforms, the more they capture and retain about you for their own proprietary use. You don’t get an option to transfer the context from SaaS ‘solutions’ like Zoom, ChatGPT or SourceGraph because in a world where your personal context is the ‘special sauce’, exporting that would be like giving away the secret sauce recipe.

Like broad-spectrum fungicides which erode the ability of the entire farming system, so does the proprietary nature of training data in AI systems. People happily working and profiting off these corporate models are then not unlike the farmers who wouldn’t serve intentionally poisoned food to their own table, but are happy to poison others’ wells and fields and livelihoods.

Nevertheless, My Bias Persisted in this AI Summary

For today, like every working day for the past four months, I’m using a homegrown process which is augmented by an LLM to take in end-of-day voice notes in order to siphon off a simplified summary. While not my own direct writing, it often captures in detail the events of the day more than I have time for, even if it comes with the occasional quirk.

Yesterday’s summary included as few bullet points on peanut butter as deer ‘deterrent’. While this is the opposite of what we use it for (attractant to cause the deer to lick the electric fence and learn to stay away), it ultimately caught the point that we were trying to deter the deer from eating the crops. Maybe some day, AI will get most of the overall purpose of all this right, even if the humans it controls used to laugh and point out it’s minor mistakes.

Consolidated Summary

Main Themes:

1. Weather Impact: Extreme heat (over 90°F) and intermittent rain affecting pest behavior, soil conditions, and work schedules.
2. Farm Tasks: Daily chores (weeding, post-harvest work, greenhouse seedling management) and adaptation to weather changes.
3. Greenhouse Operations: Seeding, transplanting, and seedling care in a controlled environment.
4. Adaptation Strategies: Adjusting work plans based on weather forecasts (e.g., prioritizing cooler days).

Activities Performed:

• Weeding: Removed weeds from lettuces and cilantro rows.
• Post-Harvest Work: Washed and packed crops.
• Greenhouse Tasks: Seeded lettuce, peppers, and bok choy; pricked out seedlings and transplanted them.
• Weather Monitoring: Noted rain’s effect on pests and soil, and tracked forecasted cooler weather.

New Things Encountered:

1. Pricking Out Seedlings: Transplanting seedlings into individual pots.
2. Greenhouse Seedling Care: Managing seedling growth in controlled environments.
3. Pest Behavior: Observing how weather influences pest activity.

Questions Raised:

1. How does pest behavior change with weather conditions?
2. Will seedlings thrive under current greenhouse conditions?
3. How can work schedules be optimized during unpredictable weather?
4. What soil management practices are effective in varying weather?

Suggested Actions:

1. Monitor Bugs: Track pest activity to adjust farming practices.
2. Prepare for Cool Days: Adjust work plans to accommodate weather changes.
3. Optimize Greenhouse Conditions: Ensure seedlings receive proper care.
4. Document Weather-Task Correlations: Analyze how weather impacts farm operations.

This summary integrates key insights from both documents, highlighting the interplay between weather, farming tasks, and greenhouse management while emphasizing adaptability and observation.

Part 1

Main Themes:

1. Weather Conditions: Extreme heat (over 90°F) with intermittent rain, impacting farming activities and pest behavior.
2. Farm Tasks: Daily chores such as weeding, planting, seedling management, and post-harvest work (washing, packing).
3. Adaptation to Weather: Adjusting work schedules and methods based on forecasted weather (e.g., moving to cooler days for labor).
4. Greenhouse Management: Seeding and transplanting crops in a controlled environment, including seed selection and care.

Activities Performed by the Intern:

1. Hoeing and Weeding:
   • Removed weeds from lettuces in the upper plot.
   • Weeded cilantro rows after a brief rain shower.
2. Post-Harvest Work:
   • Washed and packed crops.
3. Greenhouse Work:
   • Seeded lettuce, peppers, and bok choy.
   • Plucked seedlings with excessive seeds from trays (pricking out).
   • Transplanted seedlings into new trays (not potting up).
4. Weather Monitoring:
   • Noted the impact of rain on pests and soil conditions.
   • Forecasted future weather (cloudy, cooler temperatures).

New Things Not Yet Encountered:

1. Pricking Out/Transplanting: A technique to divide seedlings into new trays for growth.
2. Greenhouse Seedling Care: Managing seedlings in a controlled environment, including seed selection and tray management.
3. Pest Behavior: How rain affects bug populations and the need for future monitoring.

Questions and Future Research Areas:

1. Impact of Rain on Pests: How does the rain influence bug activity, and will this affect crop health?
2. Seedling Success: Will the new seeding method (pricking out) lead to healthier plants?
3. Weather-Driven Work Scheduling: How can farm tasks be optimized for cooler days (e.g., Wednesday-Thursday)?
4. Soil Management: Can the dry soil be improved with irrigation despite the heat?

Suggested Actions:

1. Monitor Bug Populations: Track pest activity after the rain to assess long-term impacts.
2. Prepare for Cooler Days: Plan tasks for Wednesday-Thursday (e.g., weeding, transplanting).
3. Optimize Greenhouse Conditions: Ensure proper ventilation and moisture levels for seedlings.
4. Document Weather-Task Correlation: Record how weather affects work efficiency and pest behavior for future planning.

Part 2

Main Themes:

1. Agricultural Practices: Harvesting, pruning, and preparing crops.
2. Timing and Weather Management: Impact of heat on plant growth, need for timely harvesting to prevent bolting.
3. Irrigation and Maintenance: Fixing irrigation systems, managing hose durability, and ensuring water access.
4. Plant Care: Managing plant stress (e.g., lettuce bitterness, basil bolting) and optimizing growth through pruning.
5. Seasonal Planning: Preparing for future planting (winter squash, lettuce) and adapting to seasonal challenges.

Activities Performed by the Intern:

1. Harvested chard (approx. 120 bunches).
2. Bunched and harvested basil (approx. 120–150 bunches), focusing on early growth to avoid bolting.
3. Collected younger lettuce to avoid bitterness.
4. Harvested cilantro (approx. 60–70 bunches).
5. Fixed irrigation lines by repairing holes and cracks in hoses.
6. Prepared for planting winter squash and lettuce (reverse hardening off).

New Things Not Yet Encountered:

1. Basil Bolting Management: Pricking flowering heads to prevent bolting.
2. Lettuce Bitterness: Recognizing that mature lettuce is bitter, prompting focus on younger plants.
3. Irrigation System Challenges: Dealing with hose durability and maintaining water access across fields.
4. Seasonal Planting Strategy: Preparing for winter squash and lettuce, which requires adjusting planting timelines.

Questions and Future Research Areas:

1. Optimal Harvest Timing: How to balance harvesting frequency to prevent plant stress and bolting.
2. Irrigation System Efficiency: What materials or methods can reduce hose damage and improve water distribution.
3. Plant Stress Mitigation: How to minimize heat stress on crops (e.g., basil, lettuce) without relying on refrigeration.
4. Lettuce Bitterness Management: Are there specific growing conditions or pruning techniques to reduce bitterness?
5. Seasonal Crop Adaptation: How to best prepare for winter crops while maintaining current harvest schedules.

Suggested Actions:

1. Monitor Plant Health: Regularly check for signs of bolting in basil and lettuce to adjust harvesting schedules.
2. Upgrade Irrigation Systems: Invest in durable hoses or alternative irrigation methods (e.g., drip systems) to reduce maintenance.
3. Document Harvest Timing: Track when crops reach peak quality to avoid overharvesting or underharvesting.
4. Experiment with Lettuce Varieties: Test different lettuce types or growing conditions to reduce bitterness.
5. Plan for Winter Crops: Develop a planting schedule for winter squash and lettuce, ensuring adequate water and nutrients.