Generative AI in Agriculture. Business, Land-use and Planning

Generative AI and Agriculture

The chance of Generative AI not approaching 100% adoption on farms and all businesses is zero.

This is because Generative AI is a transformative technology, able to do work faster or better than existing technologies and - here's a kicker - do work previously thought only possible by ourselves. And that is now: use of AI is enabling even more powerful AI to be developed and ultimately superintelligence - a level of intelligence beyond human intelligence.

Because it is a transformative technology the impacts that Generative AI will have in agriculture and everywhere else will tend towards the binary: like Dickens' "it was the best of times, it was the worst of times", there will be AI winners, and there will be AI losers.

In business, education, healthcare, research, agriculture and more, Generative AI solutions will be invented and adopted to perform work more efficiently and to deliver superior outcomes.

Transformative Technologies

Agricultural field irrigation systems, an Egyptian invention, were one of the first transformative technologies and ever since there has been a succession of transformative technological shifts with their occurrence becoming surprisingly frequent - surprising because we are surrounded by the outcomes of the shifts in today's now perceived as normal world. Recent examples:

Filament lightbulbs to LEDs
Celluloid film to digital photography
Typewriters to computers
Encyclopedias to Wikipedia

Perfectly illustrating the kind of binary drama that transformative technology can bring to business is Kodak.

In 1975 a Kodak research engineer called Steve Sasson invented the first digital camera. Through the next two decades Kodak secured groundbreaking digital photography patents and grew specialist digital photography markets.

But the full-frontal risk to Kodak's $16bn (1996) market-leading celluloid film business, combined with serious under-estimation of the rate of technological change, fatally delayed Kodak's entry to the consumer digital camera arena and eventually, in 2012, Kodak filed for bankruptcy. (As an aside, the chart beautifully refutes any value in automatically applying the 1995 Gartner Hype Cycle to new technologies. Successful new technologies follow a smooth, sigmoid adoption curve.)

From the Kodak perspective the 'worst of times' timescale of their 100% technological obliteration was 37 years. And yet now, here in our best of times, we regard high quality digital photography as a normal phone app.

The transformative impacts of the Generative AI revolution are set to be almost immeasurably greater than the examples above.

Why so? Because Generative AI is almost infinitely multi-functional as well as omnipresent - functioning while online and offline.

First things first, differentiation of Generative AI from other AI and non-AI will help define what the term 'Generative' AI means.

1. 'Not AI'

A good example of 'Not AI' is the autopilot used on airplanes. Here the aircraft is continuously supplying the autopilot computer with changing instrument measurements and the autopilot follows rules preset by programmers to either adjust or not adjust the flight controls accordingly. To use a computer coding analogy it's a bunch of 'If...Then...' instructions maintaining 500 knots at 40,000 feet.

2. Machine Learning (ML)

Spot weed spraying offers financial advantages to farmers of 95% to 50% savings in chemical costs over the blunt instrument approach of applying the chemical active ingredient(s) to the whole of the field. Image sensors see the upcoming ground ahead of the spray nozzles and previously taught machine learning of what weeds look like is continuously being applied to identify targets for spot applications of herbicide. Semi-analogous is a gardener hoeing her onion bed and hoeing out only weeds by recognising onion plants and recognising weeds.

3. Generative AI

The first two categories, important as they are, might be considered sophisticated, custom-built, one-trick ponies while Generative AI has literally millions of applications.

Generative AI works by having previously analysed and created patterns from data where the data are unimaginably large oodles of text, music, speech, video and images. A user's fresh input is live-analysed to enable it and the patterns to be deployed in synthesising a best fitting output.

An AI self-generated analogy - i.e. AI describing itself - is having a vast pantry of food as data; a chef knowing all of the recipes as the patterns; and then for her to be able to create existing or new recipes from the patterns as output.

McKinsey AI Results 2024

The proof of every chef's pudding is in the eating and in the case of Generative AI the eating is its adoption by businesses and organisations.

Early in 2024 McKinsey, the global management consultancy, ran this year's AI survey in a seven year linear series so the 2024 survey gives the first comparative data on changes since the first survey in the Generative AI era in 2023. The global 2024 sample comprised 1363 respondents from all world regions, 11 business sectors, all management levels and all business functions.

Some top line numbers showing the transformation (2024 v. 2023) nature of Generative AI for business were:

97% increase in use of Generative AI
66% increase in AI use in multiple functions
44% increase in adoption of all AI
+5% revenue increase by 44% of respondents
-10% cost reductions by 39% of respondents

Strongly pointing to where the maximum immediate ROA (return on adoption, since no significant cost is necessarily involved), the greatest use of Generative AI was reported as being within the marketing, product & service development and IT departments.

Further examining the 46 companies identified as reporting the greatest contributions to profits from AI, a common factor was that the top performers were deploying generative AI across more of their business functions than those in the mainstream of businesses.

Peter Gill

Managing Director

Doing business at $0.00 intelligence

How Generative AI will transform business is best demonstrated in the 2023 Harvard Business School research conducted in partnership with Boston Consulting Group and involving a large group of BCG consultants (the acme of crème de la crème business brainpower) in a trial of whether and how ChatGPT aided them in typical business consultant work tasks.

758 BCG consultants - a goodly sample size with participants from around the world - were randomly allocated to be either ChatGPT users and non-users for the trial.

Results were conclusive. The ChatGPT users increased the quality of their work by 40% and used 25% less time performing it than those without access to ChatGPT. Really significantly for the future, the greatest increases in task performances were seen amongst normally lower performing subjects.

One of the paper's co-authors, Prof. Ethan Mollick, later wrote, “I do not think enough people are considering what it means when a technology raises all workers to the top tiers of performance”. To which can be added the question:- are enough companies considering what it means when the best people are using AI and outperforming peers of equal ability but held back by not being able to make free use of AI?

The acknowledged high expertise and intelligence of the BCG people meant that the paper resulting from the research quashed qualms emanating from naysayers about the value of using AI within businesses and organisations. Businesses and organisations not investigating Generative AI solutions are, through deliberate non-engagement, courting their own Kodakesque demise.

AI and the Future of Agriculture

Farming activities can be categorised in many ways. Here, as a precursor for determining AI's agricultural roles, a thought-experiment would be to consider a dichotomous division of all agricultural work into just two sorts of activity as follows:

Moving things
Thinking about moving things

Madly reductive I know, but the purpose of our thought-experiment is to emphasise where the work is happening:-

Moving things: the first sort of activity includes, globally, all of the thousands of physical farm tasks (including the collecting of on-farm information as a 'moving things' task) such as milking dairy cows, drilling winter wheat, picking strawberries, shearing sheep, loading a grain lorry, mixing a beef cattle ration, checking crops, constructing a shed, selling a tractor and so on, almost ad infinitum. Or, the physical world of processes that when combined result in the outcome of agricultural production.

Thinking about moving things: activities of the second sort are being performed in the minds of farmers and farmer advisers to produce the short, medium and long term decisions made using human intelligence and human knowledge to keep a farm operating towards goals of profitability, business risk, sustainability, biodiversity and so on.

Now, thinking of the previous 'Not AI', ML, and Generative AI categorisations the first sort of farming activities may use 'Not AI' technology (e.g. robotic milking machines, FarmDroid seeding/weeding) and ML technology (e.g. Bayer's MagicTrap, FarmWave's Harvest Vision).

Amongst an immense multitude of other uses Generative AI will become a standard part of decision-making - the second sort of farming activity. Examples: identifying cereal diseases or weeds live in the field and recommending the best active ingredient(s) for mitigation; and helping diagnose animal health problems.

The profoundly simple reasons that Generative AI will become ubiquitous on farms are that every farmer needs information and every farmer makes decisions and employing Generative AI as on-farm co-intelligence is by far the best and most efficient means of fulfilling many tasks.

Already Generative AI is superior to websites, search engines and social media for answering questions and problem-solving. And Generative AI learns: thus as farmer useage increases AI systems expand their agricultural knowledge and all-round farming savvy. This new 'farming co-intelligence', to mint a term, will be the farmer's best friend. AI will find uses within many of the everyday and every year farm-office tasks and aiding in delivering solutions to one-off projects, for example researching all angles of a possible diversification enterprise.

Consulting farm co-intelligence will become standard farm management practice. AI is always pleased to receive probing supplementary questions or act as a supremely wise 'devil's advocate'.

The history of Generative AI and how we make use of the most amazing transformative technology ever invented has only just begun and one thing is sure - there is already much to learn and right now is the best time for everyone in business, and for every business, to be getting themselves up to speed as users of Generative AI.

Unbelievable as it may be to most people today, a point in time is coming when we won't be able to imagine fulfilling knowledge work without using Generative AI as a part of our process.