Some respondents from the larger companies and cooperatives suggest that the attitudes might be affected by the perceived inconvenience that data gathering causes.They all believe that more farmers would have a positive view on it if it was made easier for them to collect it.However, there is also a sense that the data is not used optimally, partly because it is saved in different databases that are not interconnected.The responses from the respondents indicate that data is being gathered differently depending on the agricultural sector.For instance, many respondents in the dairy section state that there is a lot of data gathered, to a high degree on an individual level, on the farm animals.In contrast, arable farmers also collect data on almost all farms, but that data is not always as detailed.An arable farmer may collect remote sensing satellite data on its farm, but sometimes not with a resolution of square meters, but rather on a field or even farm level.The inputs, i.e.the resources added to the soil, are what would be interesting for the farmer to get decision support on, if one could see a beneficial correlation between input and output.One responding farmer with previous experience from the tech industry, believes that the problem with applying AI to arable farming is the lacking volume of interconnected data.The whole data chain is not connected today, he states.In practice, the input data taken during, for example, arable seeding is not properly connected to the output of the harvest.Additionally, the insights from the harvest are not used as a decision basis for the next seeding.Thus, the data loop is not closed, which it would need to be for AI to be efficient.This data gap combined with the large amount of uncertainty factors, such as unpredictable weather, is a technical hindrance to the learning of AI models.In the field of AI and machine learning, there is an important tradeoff between bias and variance.In the interviews, the respondents had different opinions on the matter.The concept was discussed with the respondents as ‘generalizability’ and ‘precision’ instead of their technical terms.Some respondents say that precision is extremely important since a technical solution that only predicts or detects something half of the time is useless.At the same time,hydroponic grow system other respondents say that as long as the predictions are slightly better than human predictions or detections then the model can be as general as one wants.
In fact, many respondents claim that there is a much larger market for standardized models than the ones that are too adapted after local needs.There is a tendency among arable farmers and corporations that they tolerate a higher degree of generalizability while livestock farmers need more precision.A respondent in the livestock farming sector claims that a farm would never really benefit from a technical solution that could only detect rut among the animals one out of three times.Of course, many respondents bring up that there is a need for balance between generalizability and precision, and that it would be optimal if there was some degree of customizability in that aspect so that each solution can fit each farm.One key concern for the development of smart farming technologies is ownership of the data.Most smart farming systems are created as closed technological ecosystems, with limited possibilities of sharing data in between each other.This technological segregation hinders the systems to share data with each other and is thereby an obstacle to the interconnection between systems.Descending from the rivalry between the major transnational agricultural technology companies, including the quest to both pin the users to their specific technological ecosystems and avoid giving their rivals a chance to create competitive technology, this structure is difficult to change.With that said, two respondents note a tendency for transnational agricultural technology companies to move away from technology that ensnares the user to their ecosystem, to more open data flow.Such open data flow is believed to create more value for the businesses and their users.Consequently, a higher degree of data is expected to be on open standards.Even if the companies providing the technology make some progress towards open data sharing, a couple of projects are created to facilitate the data sharing compatibility.GigaCow, a research project by the agricultural university SLU on data for dairy farms, aims to enable data sharing by automatically exporting the data from different milk robots over time.Such initiatives are welcome to most farmers.However, this is a third-party work-around solution and not as straight-forward as if all machines would automatically be open for data sharing.
Some respondents lift the potential threat towards online IT systems as a risk when implementing new smart farming technology.The risk of being hacked poses a threat both to farmers and to society at large.Focusing on society at large, a respondent from a governmental agency describes cyber security as a particularly important aspect of digitalization in agriculture.This respondent believes that such a data platform probably would be classified with an extremely high security and secrecy label and be managed by the Swedish Security Service SÄPO.Therefore, this could be regarded as a clear barrier for the development process of a common data platform.Nevertheless, the respondent adds that in case of potential cyber-threats it would be better to have the data stored on a common platform than with individual farmers, since people would be managing and looking after the platform to a much higher degree than farmers currently are securing their data.Even though these issues are mostly raised by the larger organizations and authorities, the threat is also acknowledged by some farmers.They believe that connected data platforms with weak security make the farm quite vulnerable to threats.However, one farmer commented that “it is not worse than having all money in a bank account, and that I trust today.”.Other respondents, both governmental agencies and farmers, recognize the IT systems as possibly vulnerable but are not necessarily worried.Instead, they reject the belief that lacking cyber security would pose a greater threat to agriculture than to any other sector in society.When it comes to digitalization of such a fundamental societal system such as the agricultural sector, many strategic decisions are of nationwide interest.Some of the interviewed respondents from larger organizations and authorities believe that there is a wide interest that the agricultural sector becomes smarter.However, farmers are themselves accountable for making this technological transition.Two respondents argue that there is a lack of initiatives from the state or from the large organizations to drive the propagation of digitalization forward in a structured manner.One respondent, working at a governmental authority, addresses the topic of nationwide interest in digitalizing the agricultural sector , stating that AI in agriculture is a natural step moving forward.The respondent says that there are a lot of internal discussions in governmental agencies regarding if and how they should take a more active leadership role in the digitalization of Swedish agriculture.The governmental official thinks that Sweden is behind with its digital development compared to other countries with weaker economic conditions and budgets for agriculture.
A natural first step, according to this respondent, is to create a common national data platform for all agricultural data to be compiled on.Still, this respondent sees no clear political ambition driving this change, while this could speed up the digital transition tremendously.Although there is no wish to ‘force’ farmers into using agricultural technology and digitalizing their businesses, it is a likely progress if there is a nationwide and political interest in going in that direction.As in any other industry, the agricultural sector is driven by the quest for increased profit.Money is a motivator, not only for larger agricultural enterprises but also for farmers.Therefore, the general low profitability in agriculture is a major problem for farmers.Optimization plays an important role for the often unprofitable Swedish agricultural farms to be competitive on the world market.Even though there are lots of subsidies connected to food in the European agricultural system, no farmer respondents recognize any subsidies for investments in new technologies at a farm-level.Instead, the technological transition that is supposed to lead to more sustainable food production or larger output is financed by the individual farmer.different farmers have distinct economic incentives to implement smart farming technologies in their work.Generally, there is one group of farmers that have less reason to care about implementing new technologies since they will have structures in place to reach their revenue in any case.This group often owns their own property and farmland.On the other hand, there are farmers that lease their farmland and therefore constantly must become more and more effective.It is not only a matter of farm ownership though, also the size of the farm affects the probability that smart farming technologies will increase profitability.With a small farm, farmer respondents believe it is difficult to profit from smart farming techniques.A farmer with a small farm describes that he cannot afford buying new equipment, such as a new tractor, himself.Upgrading the machine park is necessary for smart farming technologies to gather enough useful data.This can be linked to the major macro trend of consolidation of farms.Basically, this means that smaller farms cannot afford to compete with the larger ones that can use their competitive advantages of being larger.There is simply not enough profit in managing most small farms, a problem which forces many farmers to merge with neighboring farms.Another trend that impacts the agricultural sector is how technologies are sold and distributed.Today, indoor garden most technology is bought as a hardware which is often a huge expense for the farmer.However, slowly things are changing.There is a transition happening towards services being bought as ‘Software as a Service’ solutions.This allows for business models in which the sold hardware is much cheaper than today or even provided at no cost, while the farmer pays a fee to subscribe for using the set of hardware and software.One respondent from an agricultural cooperative foresees that this change will have major implications and wonders whether, in ten years from now, tractors will be sold solely as a rental service instead of as a product.To enable this, an enormous amount of data will be needed.
One communicated and discussed concern about implementation of smart farming technologies is the dependency it might create towards technology.Dependency on technology refers to a system that relies on automated or semi-automated activities based on often incomprehensible software, a constant power supply or Internet-access.The system itself is not problematic to any of the respondents.However, there are some concerns regarding the cases when this type of system fails.One respondent, from an organization, states that the usefulness of the system would be compromised if the communication infrastructure would somehow break.The concern is expressed in different ways and with different urgency.Livestock farmers express their concern about this since their activities revolve around living beings, whose comfort and health rely on the technological systems continuing to operate.Also, when it comes to dependency on technology, another aspect that several respondents mention is that some practical knowledge among farmers and advisors might be forgotten.One responding farmer believes that if he applies too much technology to his farm he would risk losing some of the local, tacit knowledge of the farm.Particularly, some local variations of the farmland he finds difficult to represent correctly with data.Since there are a vast number of connected parameters affecting how a crop at a specific place will grow, he fears that a program could miss some critical aspects.This may be linked to a certain expressed mistrust towards technology, that it needs to be double checked to make sure it is doing the right thing while working autonomously.In general, there is a positive attitude towards smart farming and what it could mean, to the agricultural sector as a whole and to farmers specifically.Incorporating smart farming technologies could mean that time and costs for activities, such as irrigating and fertilizing, are reduced.Therefore, farmers can better manage their time when using well-functioning new technology.One positive side effect of this is an improved work environment for the employees.With that in mind, researcher respondent R2 states that farmers are generally bad at valuing their time spent compared to the economic return.