Productive AI solutions with proactive data management
Artificial intelligence is a beautiful name for mathematical rules that provide solutions that facilitate and develop people’s work when combined with data. Data is fed to AI models, they learn from the data, and the models produce new data. However, data is easily hidden behind the operation of artificial intelligence, so both the operation of AI and the data it uses as well as the changes that take place in them must be monitored. This need is being met internationally, for example by various countries and the EU. On April 21, 2021, the European Parliament and the Council drew up a proposal for a Regulation laying down harmonized rules on artificial intelligence – Artificial Intelligence Act.
Existing AI is based on input data from which the AI can learn. Therefore, AI has no unspecified area or activity that would be out of a human reach, so humans cannot avoid responsibility. People are needed both to understand and manage the data associated with AI and to make related decisions. It is risky to leave the responsibility to artificial intelligence itself, even though the existing narrow AI is far from super-intelligence that would be able to seize control of itself. We can happily leave these doomsday scenarios for the movie directors.
The data used in AI solutions must be managed like any other business capital. The data utilized by AI must be identified, and its management must be linked to the organization’s strategy. Clear accountability must be established for the data, which will also guide the decision-making related to the data in question. Once well established data management policies and principles have been defined, they must be put into practice at the operational level as well. The quality of data must also be developed in line with the organization’s strategy and goals. However, this cannot be done by using traditional data quality indicators, but the requirements specific to AI must be taken into consideration: sufficient data size and data storage period. All this activity must be controlled and monitored, so that AI and the data it utilizes can be used in accordance with ethical and good business practices.
AI readiness is achieved through proactive data management and data governance. The aspects of data management can be viewed as a wheel that rotates through the development of artificial intelligence, which can result in either artificial intelligence or artificial nonsense solutions, depending on how the different parts of the wheel have been managed. Is it time for your organization to develop AI capabilities or assess their maturity? To support the analysis of AI capabilities, a maturity model of data management weighted according to the development of AI has now been developed and the target maturity level of an AI-ready organization has been included, which you can get acquainted with through a Master’s thesis from a university of applied sciences here. Don’t hesitate to contact us when you need encouragement on the topic!
AIGA, (Artificial Intelligence Governance and Auditing)
This article is related to the AIGA project (Artificial Intelligence Governance and Auditing). The purpose of the AIGA project is to increase the international competitiveness and competence of Finnish companies in the reliable and controlled scaling and organizational utilization of AI. The AIGA project will study and develop AI management models and mechanisms in research and business collaboration, as well as their commercialization and export to international markets. The main sponsor of the two-year project is Business Finland.
This blog post is written by our Senior Consultant Satu Etelälahti.