Working Together on Project

Application of Machine Intelligence

for Public Good

The adoption of Machine Intelligent (MI) in the commercial sector is booming. MI technologies can also be applied by nonprofits for solving a vast number of humanitarian and environmental problems. However, doing so is a non-trivial and challenging task that requires structured analysis and sophisticated solution approach. 

MI is the term used to describe technologies that enable intelligent machines, such as, advanced analytics, data science, artificial intelligence (AI), machine learning (ML), process mining and robotic process automation (RPA). In the commercial sector, MI technologies are already being applied across a broad spectrum of industries and activities. Businesses use MI to create new products and services and to increase their operational efficiency. For example, MI is used to increase customer loyalty and revenue in online shops, social media, audio and video streaming platforms by recommending products and services to their users. MI is used in self-driving cars and smart driving assistants to improve our driving experience and transportation safety. It is at the heart of smart private assistants like Siri, Google Assistant, and Alexa, which help us to master our daily routines. Furthermore, MI helps businesses understand and optimize their processes and automate manual and repetitive tasks. It can improve manufacturing quality and identify flaws in machines before they actually break down (sometimes called “preventive maintenance”). And finally, MI helps businesses to better understand their customers via analysis of big data so that a business can better identify and react to its customers’ needs. In fact, the list of existing and potential MI applications seems to be endless. In line with this, industry analysts predict that MI technologies will soon drive over ten trillion dollars worth of the economy [1]. One need not be an industry expert to see the potential and importance of MI technologies. One only has to look at the list of the most valuable companies in the world: tech giants like Amazon, Google, Microsoft, Facebook, Tesla, etc., use MI technologies at the core of many of their products and services! Thus, MI as not a hype and not just another “fancy” technology – it is here to stay and to transform our world.

Fig1.PNG

MI for Public Good

MI opens enormous possibilities for improving our quality of life through enabling the creation of new and better commercial products and services. At the same time, humanity, particularly disadvantaged populations, continues to suffer from poverty, famine, pollution, pandemics, wars, natural disasters, a lack of access to healthcare, and discrimination. Nearly a billion people live in extreme poverty [2], almost a hundred million people had to flee from their homes in 2019 due to war or natural disaster [3], and over three million children die each year from malnutrition [4]. The good news is that MI can help solve these humanitarian challenges too!

Opportunities for serving the Public Good with MI are very impressive. On our webpage (mi4people.org), we describe various use cases of how MI can be applied for the Public Good [5], but there are many other interesting analyses available, e.g., [6], [7] and [8]. While the potential of MI for the Public Good has been studied quite a lot, it remains a challenging task to turn potential uses into practice. In this article, we focus on this challenge and propose a blueprint for mastering it.

Fig2.PNG

The Nonprofit Sector as a Major Driver of MI Application for the Public Good

 

The first question to answer is who should play a leading role in application of MI for Public Good. Some readers might see government in this role. While government support is vital and indispensable in employing MI for the Public Good, governments can sometimes be bureaucratic slow-moving machines and so may be ill-suited as the major driver of innovations. Better candidates for this mission are more agile nonprofit and non-governmental organizations (NPOs and NGOs). These organizations play a significant role in alleviating human suffering, they are experts in Public Good domains, and they represent a large part of economy – large enough part to be a major innovation driver for MI for the Public Good. From a scan of the related statistics (e.g., [9], [10], [11], and [12]) one can see that the nonprofit sector contributes between 5 and 10% to the global economy, which translates to several trillion dollars in turnover. It employs tens of millions of people involving billions of hours of activities.

Nonprofits are a large and vibrant sector whose mission is to enhance delivery of the Public Good. So, one would expect them to embrace emerging technologies, such as MI, for their operations and mission excellence. However, this seems not to be the case. In fact, surveys (e.g., [6] and [13]) indicate that only around a quarter or less of the nonprofit sector is employing any semblance of MI technologies at all, and those that do use only very rudimentary applications. In addition, the applications they do use are usually employed in the area of operational excellence (e.g., increasing output of fundraising campaigns) rather than for mission-driven applications that could enable NPOs to reach their humanitarian and environmental goals better and faster. As a result, the nonprofit sector is currently barely benefiting from the tremendous advances in MI! Why is this the case?

Root Causes for Low MI Adoption in Nonprofit Sector

 

To dive into the crux of the issue of low adoption of MI technologies among the nonprofit organizations, let us look closer at their structure and dynamics. The nonprofit sector has many players – in Germany alone there are nearly 600,000 registered NPOs [14] – but most of them are quite small and have relatively low budgets. On the other side, one of the key factors for successful MI adoption is human capital, e.g., capable Data Scientists, Machine Learning Engineers, RPA-Experts, Process Miners, etc. MI-experts are relatively rare in the market and employers compete intensively for the best talents. Bigger commercial companies with large revenues can attract technology experts more easily, because they can offer better personal compensation, the wider spread of interesting projects, and opportunities to increase MI-experts’ knowledge by working with experienced peers and mentors.

The scarcity of relevant resources and talents is not the only obstacle facing nonprofits on their journey to adopt MI technologies. NPOs are generally expected to stick very closely to their stated mission. So, they operate within the strict boundaries of regulations and sectorial norms. Trust is the primary asset for any NPO and maintaining this can often cause them to be highly conservative in their actions, avoiding experiments and – consequently – missing out on valuable innovation opportunities. NPO donors, whether individuals or foundations, are generally concerned about the reputation of organization they support. All these internal and external operational and existential constraints force NPOs to be averse to the risks associated with adopting emerging technologies, which MI is. Additionally, as some studies into the leadership styles of NPOs (e.g., [15]) indicate, NPO leaders generally have great ethical concerns, making them substantially worried about the potential unethical use of MI technologies. Of course, with the guidance and help of technology experts, it is possible to eliminate such adoption risks for MI technologies in the nonprofit sector. It is precisely this kind of experts that the sector is missing at the moment.

 

Finally, since the application of MI outside of university campuses is a relatively new phenomenon and since fundamental MI research delivers new MI technologies at an unprecedented velocity, it is extremely difficult for NPOs to keep track of these technologies and to realize their whole potential for the nonprofit sector. In fact, because of the novelty of using MI for the Public Good, it is an area which requires extensive application-oriented research and testing, before possible applications can be applied on a large scale by generally risk averse NPOs. The majority of NPOs cannot carry out such research by themselves due to a lack of financial resources and technical talent. Therefore, the gap between fundamental MI research and an actual in-field, mission-driven application of MI for the Public Good remains open.     

Fig3.PNG

Approaches to Help NPOs Adopt MI

There are several organizations whose main goal is to help nonprofits execute mission-focused MI research and projects. These organizations, however, tend to be small and they struggle to empower mission-driven MI adoption at scale. To better understand this issue, let us dive deeper into the structure and operational strategies of these initiatives. From our analysis, we identify the following major approaches:

  1. Employ own staff of problem solvers — These organizations focus on a selected small set of MI applications for the Public Good. Using their own staff, they execute a small number of projects with scopes broad and deep enough to cover conceptualization through final implementation and with the aim of delivering a high Public Good impact through them.

  2. Use volunteers — These organizations help NPOs solve specific MI related problems using volunteer workers. Thus, their projects, even though higher in volume than the previous approach, tend to have short durations and narrow scopes. These projects usually focus on solving a part of the technical problem associated with MI, for example, specific data analyses or targeted programming of a machine learning engine.

  3. Facilitate collaboration among other organizations — These initiatives bring multiple interested parties together to facilitate collaborative problem solving. They help create collaborative networks to plan and execute MI projects and may even arrange project funding.

  4. Have a mixed model – In the mixed model, organizations use some part of the strategies from all or some of the other approaches just described. Depending on the importance and complexity of a project, internal interest and capabilities, and volunteers’ availability, a suitable combination of the strategies of the other approaches are employed.

While all organizations analyzed provide added value to MI-supported delivery of the Public Good, approaches presented are not equally effective. As can be seen, the first approach can provide wholesome benefit by going deep into the issues, but it does not scale well. The second approach is more scalable than the first, but it will typically terminate the technology adoption process prematurely, i.e., before the real benefit from MI has been realized. The third approach lacks ownership and accountability and thus, although conceptually being highly scalable, in reality it does not achieve the desired scalability in terms of projects executed or impact delivered. The mixed approach, if applied carefully, has the flexibility to handle a large number of MI projects as well as drive the technology adoption deeper.

Furthermore, organizations trying to help NPOs adopt MI should appropriately engage with these NPOs throughout all stages of the innovation journey (from ideation and conceptualization to implementation and maintenance). The ability of an MI adoption-helper institution to ensure this key success factor is strongly connected to the organizational strategy and operating model of that institution. From our current analysis, we can conclude that only the first and the mixed approaches are capable of doing this.

Designing the Ideal Organization to Help MI Technology Adoption at Scale

From what has been described so far, we can make the following conclusions:

  1. MI technologies have a great potential to help NPOs better serve the Public Good;

  2. Currently only a tiny portion of NPOs use MI technologies to achieve their mission;

  3. The majority of the NPOs are quite small, preventing them from creating MI expertise in-house;

  4. Nonprofits are risk averse and cannot safely embark on an MI innovation journey without the help of a reliable, experienced, and qualified tech-partner;

  5. There is a gap between fundamental MI research/technologies and real-life applications within the nonprofit sector, resulting in an urgent need for applied research on MI for the Public Good;

  6. Organizations currently trying to help NPOs adopt MI technologies often use approaches that are not simultaneously impactful and scalable.

If we were to design an NPO that would have the ability to help the nonprofit sector adopt MI technologies in impactful and scalable ways, what would be its key characteristics? From all the facts and commentary presented so far, we can logically derive the following list of characteristics that an organization should have. Such an organization should:

  1. Be able to research, help create and maintain novel mission-driven MI applications that can be used by NPOs;

  2. Work on an open-source foundation, in order to foster adoption and improvement of mission-driven MI applications;

  3. Be easily approachable by NPOs of any size;

  4. Provide opportunities for any reasonable idea (for the use of MI) to be discussed and tested,

  5. Have capabilities and underlying processes for:

    • Analyzing and testing a large number of ideas and use cases in an efficient and effective way,

    • Extended research on and development of ideas/use cases with the largest potential for creation of scaled impact,

    • Continuous improvement and maintenance of complex MI applications;

  6. Show willingness and dedication to create long-term partnerships with NPOs;

  7. Have a strong focus on the ethical use of MI;

  8. Be backed by strong financial resources to relieve NPOs of the financial pressure of MI adoption.

The above analysis and conclusions were the motivation for us to create MI4People, a nonprofit applied research facility that takes into consideration all of the above characteristics in its organizational design and operating model. We hope that with the help of MI4People, the nonprofit sector will be able to adopt MI at scale and elevate the delivery of the Public Good to the next quantitative and qualitative level!

Fig4.PNG

Concluding Remarks

Since nonprofits are supported mostly by charitable donations and grants, and since their mission is to reduce human suffering or even death, they must achieve their missions as effectively and efficiently as possible. Hence, they should utilize any help they can get from technology, MI technologies included, to be simultaneously effective and efficient.

MI is not simply another tool that can help nonprofits (or commercial organization) become better in what they are doing; it is a major disruption driver which will change our society, the economy, and our way of life. Already now, intelligent machines influence our mobility (e.g., cars with driving-assistants), the way we communicate (e.g., social media), how we get information (e.g., AI-powered search engines), and what we consume (e.g., through personalized advertisements). And this is just the beginning! In fact, the impact of MI on humanity will probably be comparable to or even more substantial than that of the discovery of steam engine. The associated industrial revolution not only helped us produce more goods faster and cheaper, it also initiated dramatic social changes. Some of these changes were positive, like the rise of democracy and social benefit systems, but there were also negative impacts, like unplanned and chaotic urbanization, the emergence of slums, and the exploitation of workers, who were subjected to low wages and adverse work conditions.

While we still benefit from the results of the industrial revolutions (albeit more so in developed countries), our learning path towards how to utilize industrialization to benefit humanity was accompanied by wars, revolutions, exploitation, poverty, and suffering. To avoid similar scenarios for the current digital revolution, we urgently need to direct the power of MI technologies towards human flourishing and not only towards commercial profit (as is currently the case)! The nonprofit sector, as the major humanitarian ambassador, must therefore internalize MI, start employing it for their missions, and become the master of these technologies and the associated societal changes. Thus, we strongly encourage NPOs to think creatively and use the power of MI in their organization – we, at MI4People are here to help [16]!

References

 

[1] https://www.forbes.com/sites/andrewcave/2019/06/24/can-the-ai-economy-really-be-worth-150-trillion-by-2025/.

[2] https://www.un.org/en/global-issues/ending-poverty.

[3] https://www.unhcr.org/5ee200e37.pdf.

[4] https://www.worldhunger.org/world-child-hunger-facts/.

[5] https://www.mi4people.org/mi-for-public-good.

[6] https://www.mckinsey.com/featured-insights/artificial-intelligence/applying-artificial-intelligence-for-social-good.

[7] https://www.nhsx.nhs.uk/media/documents/NHSX_AI_report.pdf.

[8] https://www.pwc.de/de/nachhaltigkeit/how-ai-can-enable-a-sustainable-future.pdf.

[9] https://www.statista.com/topics/1390/nonprofit-organizations-in-the-us/.

[10] https://nccs.urban.org/project/nonprofit-sector-brief.

[11] https://prosper-strategies.com/2020-nonprofit-stats/.

[12] http://thirdsectorimpact.eu/site/assets/uploads/documentations/tsi-working-paper-no-12-size-scope-third-sector-europe/TSI-Working-Paper-12_Size-and-Scope.pdf.

[13] https://pwrdby.com/1041-2/.

[14] https://www.bertelsmann-stiftung.de/de/unsere-projekte/zivilgesellschaft-in-zahlen/projektbeschreibung.

[15] https://scholarworks.uni.edu/etd/965/

[16] We, at MI4People, are eager collaborate with you to help you understand the potential of MI and identify impactful research projects that we can undertake to find MI based solutions that help deliver Public Good better as well as assist you in meeting your organization's goals more effectively.

https://www.mi4people.org/contributetomi4people.