The challenges of integrating AI solutions in NPOs [Thesis].

Artificial intelligence is now an integral part of virtually every area of human activity. Non-profit organizations are no exception. The use of this technology is quite curious in this type of organization, given that the use of such technology is most often based on an economic objective. In some respects, this is contrary to the primary mission of NPOs, which are supposed to provide solutions to community problems; even if, to do so, this type of organization needs resources.

Integrating this technology into the management of NPOs runs the risk of distracting them from their primary mission, if the balance between this and their need for profit is not struck. Julie Soriano, intrigued by this question, proposes a work in which she questions the stakes of integrating AI solutions into NPOs - a brand new phenomenon, by the way -, with a view to raising the consequences of integrating AI on the alignment of their missions. His study is entitled: " Les enjeux de l'intégration de solutions d'intelligence artificielle au sein d'OBNL " (2018).

The answer to the main question of this study, opens with a literature review, follows with a theoretical framework and ends with research results; canvas that we will follow.

Literature review

Here, the researcher outlines the history of AI. The term was introduced by John McCarthy in 1956, during a summer seminar at Dartmouth, in order to obtain grants to enable his research group to study the potential of machines to learn and reproduce the process of human intelligence. Some thirty years later, the first chatbot was created by Joseph Weizenbaum. This was followed by the adoption of expert systems by some companies, who eventually abandoned them for lack of performance.

Since then, AI has evolved into more sophisticated forms: Machine Learning and Deep Learning. The former gives the software that uses it the ability to make predictions and self-improve from a large database, while the latter is a technology inspired by the human neural network.

Julie then presents three types of AI:

• Artificial Narrow Intelligence (ANI),
• Artificial General Intelligence (AGI), and
• Artificial Superintelligence (Asimov).

The first, described as low intelligence, rivals humans in certain domains, such as chess. AGI, on the other hand, is said to be strong intelligence. More identical to human intelligence, it is the version most eagerly awaited by researchers, companies and governments alike. And finally, the third is far superior to human intelligence (Kurzweil, 2005).

Like previous technological innovations, AI is expected to drive economic growth by reducing working hours and increasing spin-offs, despite mixed reviews. Companies seem to be particularly interested in this innovation. However, according to Olivier Ezratty, the integration of AI in business depends on three conditions:

• the definition of a problem to be solved,
• the possession of relevant data linked to the problem, and finally
• data collection and cleansing.

The introduction of AI into the world of employment has given rise to fears that certain jobs will disappear (Osborne). This idea is refuted by an OECD report, which points instead to a change in the skills and tasks of a given job. Employees can only adapt through training, which would limit the social upheavals caused by AI.

The automation of society raises a number of issues, including the gap between social classes, between industrialized and developing countries, political instability, the risk of government and corporate control...

After establishing the literature on AI, Julie Soriano lays the foundations for a review of NPOs. From this, it emerges that they are created to meet the particular needs of a population, a sign of the failure of the state, which nevertheless retains a stranglehold on this type of organization via subsidies (Defourny, 1994).

NPOs, structured in the same format as NPOs, differ from them in that the surpluses generated by their services are not distributed (Hansmann, 1980), and in the balance they strike between their mission and their search for financial resources, the mission being intrinsically linked to the organization's value. But the complexity of the NPO environment pushes them towards "entrepreneurialization and commercialization" (Julie Soriano, 2018), which could distance them from their primary social mission.

In order to verify the risk of this mission changing as a result of AI integration, the researcher develops a theoretical framework.

Theoretical framework

Julie Soriano draws on Bolman and Deal's organizational model (2017), which focuses on the structural, human, political and symbolic dimensions, in order to avoid "getting locked into the deterministic paradigm of technology and go beyond gains in organizational efficiency".

In addition to this organizational model, the researcher opts for a qualitative method to identify the actors' vision of change, the case study and the semi-directive interview. She delineates her field in Montreal, a nerve center for the development of AI technology. She targeted two NPOs operating in the artistic field. After this fieldwork, the data were examined and conclusions were drawn.

Results

After analyzing the data, we found that :

• The integration of AI in NPOs produces three consequences: the acquisition of new skills, the emergence of new needs linked to data acquisition, and a data culture.
  
  
• AI does not seem to misalign the NPO from its mission. However, the researcher sounds the alarm about the need for a direct link between the AI solution and the NPO's mission.

Illustration: mast3r - DepositPhotos

Reference

Soriano Julie, 2018, les enjeux de l'intégration de solutions d'intelligence artificielle au sein d'OBNL, HEC Montréal, Master's thesis, online
https://biblos.hec.ca/biblio/memoires/m2018a603502.pdf