Responsible Generative AI
seemingly deep cognitive functions . While the performance is impressive , given that these models show no signs of having domain basics ( understanding the science , rules , and principles ), there are lots of concerns about their widespread proliferation .
Concerns emanate from the unleashed power to do many things that educated professionals do and the fact that those tasks are done as a backbox . The following sections cover some of areas of concerns such as lack of trustworthiness , unfair impact on labor force , issues with copyrights and IP , impacts on human cognitive skills , and hence the need for appropriate regulation on the development and use of this technology .
As in the case of any fast-paced technologies , the Generative AI domain is quickly evolving and the capabilities of technology and efforts on governing the AI applications may get outdated as this paper gets published . The perspectives presented here are limited to the time at which the data is collected and documented .
2 TRUSTWORTHINESS OF GENERATIVE AI ( GENAI )
A major concern with the easy access to GenAI models is their trustworthiness . Can a user trust this AI system ? The National Institute of Standards and Technology ( NIST ) describes the essential components of AI trustworthiness as
• Validity and reliability
• Safety
• Security and resiliency
• Accountability and transparency
• Explain-ability and interpretability
• Privacy
• Fairness with mitigation of harmful bias
These models have the potential to create fake outputs , wrong answers that could be deceptive , addictive , and hence dangerous . Seemingly convincing answers to some set of questions , the power to generate texts and images often paralleling humans have created an illusion of the capabilities of this technology . There are many instances where the outputs of even welladvertised GenAI Models from industry-leading companies are found to be questionable , misleading , and often hallucinating 2 .
2 https :// www . ibm . com / topics / ai-hallucinations Journal of Innovation 17