Artificial intelligence (AI) and the fields associated with it such as machine learning (MI), natural language processing (NLP), deep learning, etc., are set to be a societal game changer. Therefore, for a sustainable future, the goal should be to incorporate the 17 development goals prescribed by the United Nations into AI with ethics at its core. However, one challenge is the opaqueness surrounding how AI processes inputs and arrives at a decision. Hence, according to Fredric Landqvist, Senior Information Strategist & Researcher with Findwise, there is a need to shift the control from the computer back to the human. This shift will enable the addition of meaning and semantics along with conceptual models.
By using open innovation—standards, models like knowledge graphs, ontologies, etc., software, and platforms—for designing AI utilities, learning can be utilized in a meaningful way. This will help shift from opaque to cognitive-informed AI and facilitate efficient communication through interoperability that can accommodate data from traditionally separate semantic domains. Similarly, open domain knowledge and datasets—linked-data—will be good platforms for continuously improving datasets within the AI loop, both in terms of refining and addressing the contextual matter and in terms of enabling improved precision and outcome.
Furthermore, there is a compelling case for building capacities at the intersection of disciplines such as computer, cognitive, data, information, and social sciences. This will help take on the challenges and opportunities within AI, by facilitating the exchange of ideas to improve value creation within various domains. In addition, building interdisciplinary capacities would herald the beginning of bringing industry, public sector, academia and society together.
AI can augment and automate utilities that benefit both the humankind and the planet. However, this requires AI artifacts to be designed according to the European Union’s draft on AI ethics. Furthermore, to enable value creation, ML, NLP and deep learning should be designed and used based on an outcome identified with the help of contextual use-cases and utilities.
Click here to read the article.
Please give your feedback on this article or share a similar story for publishing by clicking here.
