Human-in-the-Loop systems can make peace between humans and technology in sustainable investing.
We have a tendency as human beings to over-simplify things. To reduce the complexity of a problem down to something that can be solved by one or two silver bullet, one-stop-shop solutions. The same thing is happening in sustainable finance. The reliance on ESG ratings - a score encompassing and grading a company’s ESG performance - as the predominant tool for responsible and sustainable investing is endemic of this tendency.
A newer trend in this same vein of over-simplification has emerged - techno-solutionism. Techno-solutionism is the assumption that computers are superior to people and that technological solutions are superior to any other. Artificial Intelligence (AI) and Machine Learning (ML) have been particularly alluring. In sustainable finance, this is represented by the new wave of AI-driven solutions which derive insights on companies from unstructured data. Excessive reliance on these solutions in ESG risks underplaying the importance of humans in the investment process, and limits the depth and breadth of understanding that we can achieve.
By no means are we writing-off the usefulness of tech-driven solutions, however. Indeed, at Matter we believe technology has the potential to play an integral part of a holistic approach to sustainable investment that challenges that status quo, and we utilise a range of cutting-edge technologies to do so - including AI and ML. But that’s what these solutions should be - a part, not a whole.
Techno-solutionism is a widely observed phenomenon, seen, for example, when corporations and governments bank more on technological innovations than behavioural change to reduce their carbon footprint. Innovative tech solutions are sexy - far sexier than restraint, good old-fashioned manual research, or systemic restructuring.
Techno-solutionism has also been a growing concern in the field of ESG data. Since data coverage is a key issue for the industry, it was only a matter of time before ESG providers would start using AI to harness new insights from unstructured data, as a supplement to more conventional ESG data collected from annual reports and surveys. At Matter, we do the same, by using ‘natural language processing’ (NLP) to derive meaning from unstructured information about the sentiment surrounding the sustainability profile of companies and governments in online media. These are important tools to aid decision-making on non-financial themes where data is often scarce, and can create insights that would have otherwise been unimaginable without it.
"Sustainable investment at its core is about recontextualisation. It is about understanding the practical realities of how companies interact with the contexts they are in - how they rely on and impact both people and the environment."
However, trusting in AI to be the solution to trump all others has significant drawbacks:
Firstly, techno-solutionism can contribute to greater de-contextualisation in the way investors understand companies. Sustainable investment at its core is about recontextualisation. It is about understanding the practical realities of how companies interact with the contexts they are in - how they rely on and impact both people and the environment. Algorithmically generated insights from unstructured information on the internet can support this recontextualisation, but alone is insufficient. We must harness the growing range of on-the-ground, contextually grounded data on companies that is being generated by experts in the fields of civil society, academia, international organisations and beyond, on topics as diverse as human rights in supply chains to impacts on local biodiversity. And these insights should not be handled as simple input in large and obscure algorithms. Rather, they must be presented in a transparent, accessible and understandable way to decision makers.
Secondly, algorithms can create insights for you, but they can not understand meaning, importance, or value. The lack of correlation in ESG ratings highlights just how subjective a business sustainable investment can be. And while we are moving towards much needed science-based, more objective definitions of sustainability, there will always be nuances and edge cases that require investors to make decisions on what is most important and impactful. These value judgements hinge upon careful analysis of the full range of available data. Technology can help educate these decisions, but it can not make them for you.
We need to move away from being techno-solutionists towards being techno-optimists, who see the potential of technology whilst asking ‘what is the right tool for the job?’. Sometimes it will be technology, sometimes it will be humans, but ESG systems that deal with subjectivity and value judgements must combine both, and constantly reflect on the parts they play.
"Technology can help educate these decisions, but it can not make them for you..."
What we need is ‘Human-in-the-Loop’ (HitL) systems. Simply put, the idea is that ‘humans plus machines outperform humans alone or machines alone.’ These systems are designed with humans as a core part of the process. They do not flick a switch for a machine to autonomously do all of the work, by e.g. generating ESG-insights from company reports. Truly sustainable investing is too complex, too unpredictable, too full of qualitative and often sensitive social and value judgements to be solved exclusively through computation.
"Only by utilising Human in the Loop systems can we achieve an approach to sustainable investment that is both informed, and reasoned."
So, what does this look like in the case of sustainable investing?
A HitL ESG system draws on technologically derived data (e.g. data points collected, validated and interpreted automatically), combined with the growing wealth of non-technologically derived data (e.g. qualified observations from experts and verified environmental data, etc.) to provide investors with the fullest possible picture of how a company interacts with people and the planet.
At Matter, for example, we deploy AI systems to evaluate sustainability-related sentiment in the news, but we also use the technology to collect existing and validated insights from sustainability experts, and to structure and bring this deep and diverse data together in the purest and most transparent possible way. Put more simply, we see it as the responsibility of our technology platform to bring deep insights from humans together, rather than to replace humans in the evaluation of the sustainability of corporations and states.
Technology, when used in this way, can augment the capabilities of humans by facilitating and empowering human decision-making. Technology serves as a tool and a bridge. It acts as part of a wider ecosystem of knowledge and solutions, rather than the system itself. It is a means, but not an end.
The emergence of new and transformative technologies, therefore, should free up ESG professionals to focus on doing what technology cannot, which is deriving meaning and value, and ultimately, making the tough decisions.
Only by utilising Human in the Loop systems can we achieve an approach to sustainable investment that is both informed, and reasoned.
We have a tendency as human beings to over-simplify things. To reduce the complexity of a problem down to something that can be solved by one or two silver bullet, one-stop-shop solutions. The same thing is happening in sustainable finance. The reliance on ESG ratings - a score encompassing and grading a company’s ESG performance - as the predominant tool for responsible and sustainable investing is endemic of this tendency.
A newer trend in this same vein of over-simplification has emerged - techno-solutionism. Techno-solutionism is the assumption that computers are superior to people and that technological solutions are superior to any other. Artificial Intelligence (AI) and Machine Learning (ML) have been particularly alluring. In sustainable finance, this is represented by the new wave of AI-driven solutions which derive insights on companies from unstructured data. Excessive reliance on these solutions in ESG risks underplaying the importance of humans in the investment process, and limits the depth and breadth of understanding that we can achieve.
By no means are we writing-off the usefulness of tech-driven solutions, however. Indeed, at Matter we believe technology has the potential to play an integral part of a holistic approach to sustainable investment that challenges that status quo, and we utilise a range of cutting-edge technologies to do so - including AI and ML. But that’s what these solutions should be - a part, not a whole.
Techno-solutionism is a widely observed phenomenon, seen, for example, when corporations and governments bank more on technological innovations than behavioural change to reduce their carbon footprint. Innovative tech solutions are sexy - far sexier than restraint, good old-fashioned manual research, or systemic restructuring.
Techno-solutionism has also been a growing concern in the field of ESG data. Since data coverage is a key issue for the industry, it was only a matter of time before ESG providers would start using AI to harness new insights from unstructured data, as a supplement to more conventional ESG data collected from annual reports and surveys. At Matter, we do the same, by using ‘natural language processing’ (NLP) to derive meaning from unstructured information about the sentiment surrounding the sustainability profile of companies and governments in online media. These are important tools to aid decision-making on non-financial themes where data is often scarce, and can create insights that would have otherwise been unimaginable without it.
"Sustainable investment at its core is about recontextualisation. It is about understanding the practical realities of how companies interact with the contexts they are in - how they rely on and impact both people and the environment."
However, trusting in AI to be the solution to trump all others has significant drawbacks:
Firstly, techno-solutionism can contribute to greater de-contextualisation in the way investors understand companies. Sustainable investment at its core is about recontextualisation. It is about understanding the practical realities of how companies interact with the contexts they are in - how they rely on and impact both people and the environment. Algorithmically generated insights from unstructured information on the internet can support this recontextualisation, but alone is insufficient. We must harness the growing range of on-the-ground, contextually grounded data on companies that is being generated by experts in the fields of civil society, academia, international organisations and beyond, on topics as diverse as human rights in supply chains to impacts on local biodiversity. And these insights should not be handled as simple input in large and obscure algorithms. Rather, they must be presented in a transparent, accessible and understandable way to decision makers.
Secondly, algorithms can create insights for you, but they can not understand meaning, importance, or value. The lack of correlation in ESG ratings highlights just how subjective a business sustainable investment can be. And while we are moving towards much needed science-based, more objective definitions of sustainability, there will always be nuances and edge cases that require investors to make decisions on what is most important and impactful. These value judgements hinge upon careful analysis of the full range of available data. Technology can help educate these decisions, but it can not make them for you.
We need to move away from being techno-solutionists towards being techno-optimists, who see the potential of technology whilst asking ‘what is the right tool for the job?’. Sometimes it will be technology, sometimes it will be humans, but ESG systems that deal with subjectivity and value judgements must combine both, and constantly reflect on the parts they play.
"Technology can help educate these decisions, but it can not make them for you..."
What we need is ‘Human-in-the-Loop’ (HitL) systems. Simply put, the idea is that ‘humans plus machines outperform humans alone or machines alone.’ These systems are designed with humans as a core part of the process. They do not flick a switch for a machine to autonomously do all of the work, by e.g. generating ESG-insights from company reports. Truly sustainable investing is too complex, too unpredictable, too full of qualitative and often sensitive social and value judgements to be solved exclusively through computation.
"Only by utilising Human in the Loop systems can we achieve an approach to sustainable investment that is both informed, and reasoned."
So, what does this look like in the case of sustainable investing?
A HitL ESG system draws on technologically derived data (e.g. data points collected, validated and interpreted automatically), combined with the growing wealth of non-technologically derived data (e.g. qualified observations from experts and verified environmental data, etc.) to provide investors with the fullest possible picture of how a company interacts with people and the planet.
At Matter, for example, we deploy AI systems to evaluate sustainability-related sentiment in the news, but we also use the technology to collect existing and validated insights from sustainability experts, and to structure and bring this deep and diverse data together in the purest and most transparent possible way. Put more simply, we see it as the responsibility of our technology platform to bring deep insights from humans together, rather than to replace humans in the evaluation of the sustainability of corporations and states.
Technology, when used in this way, can augment the capabilities of humans by facilitating and empowering human decision-making. Technology serves as a tool and a bridge. It acts as part of a wider ecosystem of knowledge and solutions, rather than the system itself. It is a means, but not an end.
The emergence of new and transformative technologies, therefore, should free up ESG professionals to focus on doing what technology cannot, which is deriving meaning and value, and ultimately, making the tough decisions.
Only by utilising Human in the Loop systems can we achieve an approach to sustainable investment that is both informed, and reasoned.
