The Future of AI
By Sam Watson Jones, Co-founder at Small Robot Co
How the influence of Artificial Intelligence is growing in farming
What is the future potential for Artificial Intelligence (AI) in farming? And where will it lead us in the next two to three years?
Firstly, I’m going to illustrate how the influence of AI has expanded over the last two to three years, before looking ahead at likely developments over the next two to three years.
The last 2-3 years – lessons from medicine and healthcare
I would argue that the healthcare sector has a more innovative mindset than agriculture and R&D in this sector is often better funded. This is particularly apparent in the US where there are huge sums of money to be made from breakthrough healthcare technologies and an ecosystem in place to support these innovations. It can also be a useful tracker for the sorts of developments which may appear further down the line in agriculture.
That’s why I take a keen interest in technology developments in medicine and healthcare. The healthcare industry, like agriculture, is dealing with living things and therefore a huge number of uncontrollable and sometimes unknown variables.
Over the last 2-3 years months there has been a proliferation in the use of AI to process and analyse test results, monitor patient progress and suggest options for treatment. Way back in 2019 The Lancet, a medical journal, reported that AI was already delivering equivalent diagnostic performance to human medical professionals when identifying diseases via medical imaging. Over the three years since then, there are numerous studies showing that AI algorithms surpass human doctors in diagnosing a range of diseases. In particular, AI is better at identifying the more rare diseases, which human biases are more prone to overlook. Recently the US had its first case in which a doctor was sued for not using AI to support a diagnosis when AI support was available.
In recent years, AI has played a crucial role within the modern healthcare setting.
Healthcare has moved from understanding the promise of AI to a position in which, in many cases, it feels unusual for a doctor to take a decision without the support of AI. A wider range of surgeries are also being carried out with AI support or with robots.
It seems logical that crop farming will follow a similar trajectory.
Sooner than we think, it will feel unusual to take any decision about what should happen in the field without consulting Artificial Intelligence. Sooner than we think, we will look back on the era of blanket applications of chemicals and fertilisers as a bizarre time in our past, and every action that we take in our field will be highly precise, down to the Per Plant level.
The next 2-3 years
One of the biggest blockers to AI transforming crop production has been the lack of granular, digitised datasets. Those that have existed up to now have been too high level to have a meaningful impact.
The years 2021 and 2022 have seen a shift towards Artificial Narrow Intelligence for agriculture. This is where machines are getting increasingly smart on a specific set of tasks, such as autonomously detecting weeds, pests or diseases in crops. Lots of work that is going on in these areas in small scale trials and pilot projects will burst into the mainstream.
Farmers will start to get used to having their decision making supported by machines which are autonomously gathering, analysing and presenting data to them. It will begin to become normal for farmers to use AI to blend layers of data together to make better decisions.
AI will soon give farmers the confidence to make better decisions on their own farms.
Farming will remain a relationship business, but the most forward-thinking farmers will begin to realise that one of their most important relationships will be with the technologies they’re using to make day to day decisions on farm. It will become more important that the farmer is able to ask the right questions than know the answers to all the questions. This is a fundamental shift in the mindset of the industry, which prides itself on individual expertise at present.
Towards the end of this period, we will begin to see the first signs of system change in farming as algorithms begin to discover patterns in agriculture previously hidden from human operators.
Change to thinking
This could change our thinking about which crops to grow in what areas, how to minimise the risk of pest damage and which weeds we need to focus our attention on first. Answers could also come on how to reduce chemical usage, increase and measure carbon sequestration capacity, maximise biodiversity in arable fields and many more areas.
In three years’ time, AI powered robots will have a strong foothold in the arable market - somewhere between 5-10%. It will still only be the early adopters who are using this technology across a broad acreage.
Most farmers are likely to be operating a farm that looks very similar to the one they’re farming today. But within this timeframe the broader industry will be starting to take decisions with the support of AI tools such as Small Robot Company’s AI-driven Advice Engine, even if they’re not yet using robotics.
The key shift over the next three years will be around the credibility of these new technologies as a driving force for farming’s future. Questions about their technical feasibility will no longer be valid. The conversation will have shifted to one of prioritisation and the most appropriate business models for enabling these technologies to become ubiquitous throughout the industry.
Artificial Intelligence will transform the farming industry. Robots are the tools to enable this transformation at scale.
Examples of Per Plant Farming
Per Plant Farming is the ability to gather data on every single plant in a field, and to take action at that same level of granularity.
This spring, Small Robot Company conducted a trial with Tuckwells - an agricultural equipment dealer in Suffolk. We scanned a field and counted 1700 weed plants. We converted this into a shape file and transferred this to a conventional farm sprayer. Using individual nozzle control we went through and applied the herbicide on the field. Rather than applying it to 100% of the field we only needed to apply it to 3% - a saving of 97% of herbicide.
As well as the direct savings of herbicide, this informed spot spray enables the farmer to know exactly how much chemical to load into their sprayer before they enter the field, so there is no need to dispose of unused chemical.
We believe that the same approach, using very high-resolution imagery which instructs the machine to take more accurate action - can be transferred to fungicides, pests and nutrition maps. In some instances, the action will be carried out by existing farm machinery, in others farmers will use precision spraying robots to achieve an even higher level of accuracy. We believe it is the start of a transformative approach to farming – the fourth agricultural revolution where Per Plant Farming will have a key role.
“Per Plant Farming is the ability to gather data on every single plant in a field, and to take action at that same level of granularity”.
