If our society would be a research task, the hypothesis could be postulated as
AI + BUSINESS = COMPETITIVE ADVANTAGE
In order to approve or disapprove the afore mentioned hypothesis, one could use research methods of various nature.
Feasibility studies
Any company verging on any AI solution should determine the viability of this solution – ensuring the project is legally and technically feasible as well as economically justifiable. Before proceeding, it is important to determine if the project is worth the investment. The economic effect from an AI intervention could be measured in terms of operational efficiency, system reliability and stakeholder satisfaction. Such studies entail both scientific and practical aspects.
Practically oriented studies help companies to make informed decisions. Companies need to get things right the first time, especially when committing large resources in terms of budget and personnel. Conducting a feasibility study identifies new opportunities and helps to narrow down business alternatives. There should always be a valid reason for undertaking an AI project – which is an intervention into historical way of running things. These studies focus on constraints, both internal and external. Internal constraints include company’s resources (budget, personnel), operations and technical readiness. External constraints look onto outside connections, such as legal framework, binding regulations, logistics and the environment.
Technical feasibility studies assess the technical resources needed and available tot he company. Technical feasibility shows, if the company is technically ready to implement an AI system and to interconnect it with all the necessary processes. Technical feasibility includes both technical resources capacity and the technical team capabilities.
Operational feasibility studies assess the benefits a company gains from completing an AI introduction into the busines processes. This reveals, if and to what extent the company’s operational requirements and expectations are met by the intervention (AI installation).
Economic feasibility studies look at financial gains from a planned AI system. This entails cost and benefit analysis which determine if the costs are viable and the benefits outweigh potential loss.
Legal feasibility studies look into the planned AI system’s potential conflicts with legal requirements, such as data protection and other laws. The legal regulation of AI is currently in rapid development. The E.U., U.S. and U.K. have been forerunners in tackling legal regulation of AI technologies. Understandably, the companies want to be certain of any current AI solutions or new investments into such conforming with both the current and future legislation.
Domain knowledge Algorithmic thinking
↓ ↓
Development of effective business AI solutions
↓
Feasibility studies
↓
Intervention studies
↓
Validated AI solutions
Why do AI and business researchers make a good team? The above depicts a full pipeline of business AI research.
Intervention studies
Intervention could be described also as testing of an AI solution under real life conditions. Interventions are studies that test the effects of changes in a business’s operations. An intervention study tests a treatment or program. The control group is not given the treatment, but rather serves as a baseline for comparison. There are many different types of interventions, but they all involve an element of control: one group is given something while another is not, or one group experiences something different than another.
Intervention studies are used in business and industry to evaluate new processes, products, advertising campaigns, and other business processes that are designed to affect business efficiency , consumer behavior or other performance metrics.
The purpose of an intervention study is to determine whether a new AI solution will have a positive impact on the business operations. Intervention studies can be used by businesses as a way to test out new methods of work before they are rolled out. Companies may also use them to test existing products or services that they want to improve upon using AI.
Intervention studies could be conducted for various reasons, including:
- To test whether an AI solution will work in the real conditions;
- To determine how effective an existing (non-AI) business process is;
- To see if there is room for improvement within an existing product or service, using AI; and
- To ensure customer satisfaction levels are high enough for certain products or services.
EXAMPLE 1 A company is testing out a new marketing strategy, where you might run an intervention study where you only use your new AI generated content for one month. The control group would be given no information about your new marketing plan at all; they'd just go about their daily lives as usual. The intervention group would receive all the information about your new marketing plan and then be encouraged to engage with it however they could. Then compare it to how things were before that month, when using human copywriters. In an intervention study, researchers change one part of their business and keep everything else constant so they can measure the change.
EXAMPLE 2 In another example, a company wanted to know whether or not a certain AI driven management method would help workers to work more effectively, one could conduct an intervention study. The company would have one group of workers using the new management method, while other workers do not. Then their intervention test results are compared to determine which group performed better on the tests given after the intervention was administered.
EXAMPLE 1 A study might look at the effect of adding self-service cashiers to a store, where AI compensates for the lack of human cashier. Relevant research questions arise. Does this increase sales? Or does it make customers wait longer in line?
We collaborate with businesses
Besides making research, we are also offering our expertise also for business people, entrepreneurs and startups who are interested in learning the concepts or implementing of Artificial Intelligence with Machine Learning, Predictive Analysis, Forecasting, Testing and Validation.