I'd suggest that the challenges are similar to other use cases for AI - lack of specific-enough data, hallucinations, etc.
Specifically for risk analysis, to make the risks meaningful you'd need to provide sufficient details about the project. Doing so might break your organization's security or AI-specific policies unless you have a private instance of a given AI tool which this data can be safely provided to.
PMO Leader | Speaker & Mentor | Content Leader – PMOGA Latin America
Hub| Catholic University of UruguayMontevideo, Montevideo, Uruguay
AI-generated risk analysis can be useful, but its reliability depends on the quality of the data. The data must be accurate, relevant and free of bias, as any errors or inconsistencies in the data will directly affect the results of the analysis.
AI processes information, identifies patterns and generates predictions based on the data provided. If the data is reliable, AI can provide robust analysis. However, if the data is incomplete, outdated or incorrect, the conclusions will be incomplete, outdated or incorrect.
Therefore, the key is not just the technology, but ensuring that the data used meets the quality standards needed to support informed decisions. Saving Changes...
Project Manager| AWR Development (BD) Ltd. Cox's Bazer , Bangladesh
Hi Tariel Barragan,
AI-generated risk analysis can be useful, but its reliability depends on data quality and algorithms.
However, challenges include data quality, the need for human oversight, and keeping models updated.
Have you had any experiences with AI in risk analysis? What challenges did you encounter?
Best,
Golam Rob Saving Changes...
Sergio Luis ConteHelping to create solutions for everyone| Worldwide based OrganizationsBuenos Aires, Argentina
Yes, I am using from long time ago. For example, lot of well known tools to make things like Montercarlo simulation have AI embeded from more than 30 years ago. When you use AI, any type of AI, the key success factor is to understand that one of the basement of AI is the concept "human in the loop". AI will give al least 3 results with probabilities associated to it. Then you have to choose the best way based on the results. But the choice is yours. Regarding of reliability, you can start by calculating the confidence rate based on the available information to give as input to the AI component. Usually data engineers or knowledge engineers are on charge of that. Saving Changes...
Community Champion
