Experts discuss the hurdles organisations face in scaling AI initiatives, highlighting the necessity of unified data and governance frameworks.
Recent discourse surrounding the integration of artificial intelligence (AI) within business practices has highlighted numerous obstacles that organisations face when trying to scale their AI initiatives to full production. Key issues identified include a lack of alignment within businesses, uncertainty regarding the trustworthiness of AI models, and ongoing regulatory concerns. According to experts, unified data and AI governance are deemed essential for navigating these challenges effectively.
During a recent webinar hosted by SAPinsider in collaboration with Collibra, Kat Cheng, Senior Director of Product Marketing for SAP Data & Analytics, stated, “Managing hybrid and multi-cloud environments, increasing data volumes, and the demand for trusted data are just some of the challenges faced by data stewards in large enterprises.” Cheng elaborated that the data needed for effective AI applications is derived from a mixture of existing and newly developed applications, presenting significant hurdles for organisations.
Cheng pointed to SAP Datasphere as a key solution, which offers a robust data foundation necessary for managing both SAP and non-SAP analytic applications. This platform supports a fabric architecture aiming for seamless and scalable access to business data while ensuring that critical semantics and business context are preserved.
SAP Datasphere incorporates features like data integration, data virtualisation, and self-service data access. Cheng noted that it is capable of harvesting detailed metadata lineages from various SAP sources, including analytics and cloud platforms, presenting this information comprehensively within SAP Datasphere. This level of integration is critical in providing clear visibility across an organisation’s data landscape.
AI governance has emerged as a natural extension of data governance, particularly when multiple stakeholders collaborate towards a common goal. Mike Robertson, Vice President of Field Alliances at Collibra, addressed the current complexities faced by data leaders, stating, “Some of the challenges that data leaders are facing with AI now is that there is a wider variety of semi-structured and unstructured data.” He emphasised the importance of utilising reliable data to bolster machine learning and AI initiatives while ensuring that all data is traceable to support various AI projects.
Robertson also described the fundamental aim behind the native integration of Collibra with SAP Datasphere: “The goal was to create a unified governance framework that supports both data and AI initiatives, ensuring compliance with regulations and mitigating risks.” This structured approach is pivotal for organisations wishing to navigate the intricate regulatory landscape while managing data-driven initiatives.
Additionally, Vasiliki Nikolopoulou, Principal Integrations Architect at Collibra, demonstrated the benefits users receive from the integration, which allows them to explore AI models and relevant use cases that align with business objectives, such as sales forecasting. “The integration allows users to filter and find relevant AI use cases, providing a business context around AI models and their associated data,” she noted, underlining the need for risk assessments and compliance tracking in the realm of AI governance.
These discussions underscore the significance of data quality in enhancing the predictive capabilities of AI technologies. By harnessing unified data and AI governance, organisations can foster a comprehensive understanding of their data environments while mitigating the associated risks and compliance requirements. This approach will be crucial as businesses continue to explore and implement AI technologies as a foundational layer in their operations.
Source: Noah Wire Services
- https://blog.getaura.ai/ai-integration-challenges – This article supports the claim that many companies struggle to achieve scalable value from AI integration and face difficulties integrating AI with existing systems, highlighting the need for strategic planning, data readiness, and cultural adaptation.
- https://blog.getaura.ai/ai-integration-challenges – It corroborates the idea that AI integration challenges are both technical and human-centric, requiring focus on data governance, workforce development, and cultural adaptation.
- https://www.ibm.com/think/topics/trustworthy-ai – This article discusses the importance of trustworthy AI, including the need for explainable, fair, interpretable, robust, transparent, safe, and secure AI systems to mitigate risks and build trust.
- https://www.ibm.com/think/topics/trustworthy-ai – It highlights the challenges associated with the lack of transparency in AI models and the need for continuous monitoring and AI governance frameworks to ensure trustworthiness.
- https://www.sightfull.com/resources/blogs/ai-impact-on-business-analytics/ – This article addresses the challenge of integrating business data into AI-based analytics platforms, emphasizing the difficulties with diverse databases and the lack of transparency in AI tools.
- https://www.sightfull.com/resources/blogs/ai-impact-on-business-analytics/ – It supports the notion that the lack of transparency in AI models is a significant drawback for businesses requiring in-depth data scrutiny and audit trails.
- https://www.frontiersin.org/journals/big-data/articles/10.3389/fdata.2024.1381163/full – This research paper discusses the critical gaps in AI risk management frameworks, including the neglect of human factors and the importance of addressing human biases and errors in AI development.
- https://www.frontiersin.org/journals/big-data/articles/10.3389/fdata.2024.1381163/full – It underscores the need for transparent communication about how AI operates and its limitations to enhance trust in AI systems.
- https://blog.getaura.ai/ai-integration-challenges – This article emphasizes the importance of starting small, prioritizing data readiness, and committing to continuous learning to overcome AI integration challenges.
- https://www.ibm.com/think/topics/trustworthy-ai – It mentions the role of various frameworks and guidelines, such as those from NIST and the European Commission, in ensuring trustworthy AI.
- https://www.frontiersin.org/journals/big-data/articles/10.3389/fdata.2024.1381163/full – This paper highlights the importance of a comprehensive approach to AI trustworthiness, combining technical and social mitigation measures to address various risks associated with AI integration.
