Technology has been advancing faster than we ever thought possible. Robotics, large language models (LLM), artificial intelligence (AI), and automation are a few of the gains from the past years that have helped us work more efficiently and with greater precision. It is now possible to create text, summarise documents, and create photos or videos within seconds. ChatGPT and the capabilities of generative AI are fascinating, to say the least. Why is it then, that so little of this advancement has made its way into the world of impact assessments?
Due to the travel restrictions, I remember 2020 as a defining year for experimenting with satellite technology for construction progress monitoring. It was mind-blowing to me that the clear satellite images could even show us the roofing material so we could verify resettlement compensation from our desks in Beijing. Environmental colleagues have been experimenting with various monitoring and survey tools for biodiversity aspects, including bird monitoring for shutdown on demand for wind projects.
The world of impact assessment has also shown interest in using technology to make life easier. In a recent paper on using AI, the researchers wondered whether AI could be used to review the literature on the use of AI in EIA. The conclusion is that incorporating modern technology into the way we work is probably inevitable, and hopefully, this will be done with transparency and an ethical approach.
The limitations and concerns around AI for impact assessments can be summarised by the lack of citations and the question of whether answers are reliable and ‘true.’ There seem to be concerns about the transparency of what data is provided to the models to ‘learn’ from. This is especially true for sentiment analysis and predictions based on historical data. Some applications provide references, but as the research paper above suggests, these are generic and inappropriate for our E&S assessments. It is indeed challenging to replace a human E&S expert with a machine, especially when the context is not provided for the machine to understand technical questions.
I am still hopeful that advancements in technology will translate into progress for us. As I wrote in my book ‘What is a Social Impact?’, I am dreaming of a world where the challenges I face on a daily basis are addressed:
Large amount of decentralised data
If only there was a centralised database where I could access data from national statistical offices, past projects from different lenders, and local and international legislation on different E&S topics. Now, I must spend a tremendous amount of time searching across multiple sites in different languages.
The language barrier
I am surely not the only one who works in countries where we do not speak the local language. Online translators are -somewhat- helpful, but I still struggle with browsing national statistics or local legislation. My current options are using my phone’s translation program or trying to copy and paste text into the browser-based translation. Surely, this should be a solved problem!
Institutional knowledge gone with each retirement or sharepoint upgrade
Misplacing documents, colleagues who don’t file, and people leaving organisations without proper handover notes are just a few of the issues that lead to a significant loss of institutional knowledge. I can not count the times I have been looking for precedents for ESAP items, project management plans, or mitigation measures for unique issues. The best course of action is also to ask that one person who has been with the company for a long time and remembers the agreed ‘best practice’. The other issue is the upgrade of the system where a large number of documents should seamlessly migrate to their new home – except when they don’t! I wish there was a solution to save time by figuring out how to make this knowledge easily accessible!
Harmonising approaches for risk mitigation – same ESAP item for similar problems
Clients are getting smarter, and they recently called us out for requiring different things for a similar impact on two of their projects. Harmonizing institutional approaches is indeed challenging when clients work across regions with different members of the E&S teams assigned. While complete harmonization of approaches might be challenging—client willingness, project-specific needs, and contextual risks differ—there is still a lot that we could learn from having access to similar projects from the past.
As the research states, it is not a question of if but when we will incorporate technology into the way we do E&S performance management. I am personally very interested in how to use technology to make certain tasks and workflows more efficient while relying on our human brains, expert knowledge, judgment, and experience to draw the final conclusions for each project. I believe that AI can be used in a way that provides transparency to the ever-growing volume of data that we are faced with and can deliver accurate responses with verifiable references to our research queries. The question is when the industry will be ready to embrace this!
Comments