AI in contaminated land
AI is the current holy grail in tech circles, appearing in virtually every app or web service that promises to ‘save time and improve on workload’. But, in essence, what’s happened lately isn’t so much a revelation as an evolution of previous AI capabilities, one which now allows for greater machine understanding and interaction with the real prime movers; humans.
So yes, most of us will still have our jobs to go to. But, just as car manufacturers were forced to learn the assembly line in the industrial revolution, and engineers software programming after the introduction of modern computing, so too will AI change the way we currently work.
‘We can’t manage what we can’t measure’.
This statement came from a UN environmental program article amidst the excitement of new AI tools late last November. The article goes on to quote David Jenson, coordinator of the United Nations Environment Program's (UNEP's) Digital Transformation sub-programme, as he talks about what it is AI is actually doing for humanity now.
In essence, Jenson says, AI performs jobs that human minds were typically meant to do. Just as an assembly line robot ‘mimics’ the actions of human engineers, AI mimics the capacity for analysis that we humans have. Where AI offers greater value is in the scale at which it can perform this monitoring and analysis; reviewing entire satellite transmissions of global emissions, for example. Which is exactly what the UN’s WESR (World Environment Situation Room) is currenty doing.
Who monitors the monitors?
However, the whole point of greater monitorisation and analysis is to inform decision-making on a human level. The greater the analysis, the more accurate are the predictions that users can make and the more definite are the actions that organisations and governments can take. Decision making, and interpreting the datasets into real-world actionables, is still something that requires human attention.
In short, someone still needs to monitor and make sense of the AIs.
How this impacts land contamination.
The increase in data analysis potential from AI has opened the way for numerous possible studies and land contaminant concerns within New Zealand and abroad.
The AI forum is one such official institute that has published a report on the various uses of AI driven data sets, for the purpose of monitoring biodiversity, land use, climate change, water ways and resource allocation. As of yet, the application of AI between different organisations and disciplines has been inconsistent, with the practicable technology too new a development to factor into governmental plans. The use of AI within the Three Waters Legislation, for example, is still a topic of debate at the time of writing, as local councils struggle to find ways to regulate water usage.
However, AI is the subject of numerous critical studies, particularly within waste disposal and reduction methods. It’s also been tested over the last decade in a official capacity to monitor ‘spill events’ and waste water management in countries such as the UK and India, with larger populations relying on often outdated water supplies. The same methods are being utilised to monitor soil quality and in agricultural development also, creating datasets that predict soil quality and fertility.
Why isn’t the change over happening faster?
Machine learning is limited by a number of factors, the main one of which isn’t easily solvable. Because, essentially, AI is a big brain without any direct connection to arms and legs.
What we mean by this is that AI, particularly within land contamination, is dependent on data collected by humans. So while the speed of processing and analysing has increased, as has the sophistication of the answers AI can give, the amount of viable, up-to-date information we can collect is a much slower process. And, unlike data analysis, the amount to which machines can take over the workload of human data collection is currently very limited.
Where to from here?
AI, or at least the current AI we are familiar with, is totally reliant on the information we give it. Over time, as AI is open to more datasets, its ability to learn will grow too.
For those at the forefront of land contamination, the change will be in the way reporting tools are able to make sense of datasets. But in the early stages of development, and with limited datasets, the ability of AI to measure and predict faults may vary. For contamination reports that rely on historic information too, this requires more updates and the integration of modern monitoring and reporting methods with an AI.
Will AI eventually overrule human decision making?
The best guess is that it will augment it, not replace it. So, far from being out of jobs, land contamination specialists will require a new set of skills — interpreting AI output and using the faster reporting metrics to make their own decisions. Deciding where to invest funding and whether a decision is bad long-term versus the need for development is still very much a decision that humans, and in particular land contamination specialists, will have to make for themselves.
- Ali, Owais. “Artificial Intelligence (AI) in Soil Quality Monitoring” on AZO Robotics. Date Published: 1st March, 2022. Site Link: https://bit.ly/40KQ9k1.
- Cremer, David De, Kasparov, Garry. “AI Should Augment Human Intelligence, Not Replace It” on Harvard Business Review. Date Published: 18th March, 2021. Site Link: https://bit.ly/422rj01.
- Fawcett, Sephi. “How is AI revolutionising the contamination sector?” on contamination expo. Date Published: 30th March, 2021. Site Link: https://bit.ly/3Ay6yO7.
- “How artificial intelligence is helping tackle environmental challenges” on UN environmental program. Date Published: 7th November, 2022. Site Link: https://bit.ly/3LajONN.
- Luhrs, Jim. “Three Waters — Three solutions” on Medium. Date Published: 10th Februrary, 2023. Site Link: https://bit.ly/41FL3ql.
- Moradi 1, Mohammad Hossein, Sohani, Ali, Zabihigivi, Mitra, Wagner, Uwe, Koch, Thomas, Sayyaadi, Hoseyn. “Chapter 12 - Machine learning and artifical intelligence application in land pollution research” in Current Trends and Advances in Computer-Aided Intelligent Environmental Data Engineering, Intelligent Data-Centric Systems, pp. 273-296. Date Published: 2022.
Typically used in various geological and environmental fields to identify and communicate problems and observations, GIS maps show trends, help detect patterns, and help people respond to ongoing changes within an observable area.
Unlike a standard report, Datanest is designed primarily to capture, interrogate, manage and deliver data, meaning it can consider more complex data patterns than standard reporting tools.
The greatest issue that software designers can’t (completely) account for is the physical limitations of phones themselves and how to harness the benefits of new software in challenging outdoor environments and situations.