Navigating AI Success: Overcoming Leadership Challenges in Large Organizations

Welcome to the AI Engineering Summit, where we're tackling the challenges of leadership, strategy, and technology in generative AI projects, featuring co-presenters Jonathan and Stephen from Fiser.

1. The speakers are here to discuss leadership and how to apply concepts in practice, specifically talking about strategy, technology, and generative AI (Gen).
2. Gartner predicts that 30% of Gen projects will be abandoned by 2025, but many people in the audience have experienced challenges with Gen projects.
3. The speakers emphasize the need for a clear vision and understanding of Gen when presenting it to executives who might not be technically savvy.
4. Jonathan shares his experience implementing Gen in a Life Sciences company, focusing on technology transfer from lab-scale to industrial-scale drug development.
5. A key challenge is that the average tenure of manufacturing workers has significantly decreased, and much expertise will soon retire due to retiring baby boomers.
6. To address this issue, Jonathan's team used Gen AI to understand and manage the millions of documents and notes generated during drug development.
7. They broke down documents into chunks and loaded them into a graph, structuring the information in a way that made it more accessible and usable.
8. By using similarity search and refining how they stored and managed chunks, they were able to improve their system over time.
9. The main business challenge is convincing teams to move from existing systems to more expensive Gen architectures, which require an R&D investment.
10. To do this, it's important to understand the executive's goals and show how Gen can help achieve those objectives.
11. In large organizations, messages from executives trickle down through various levels, with each level having specific concerns (e.g., cost savings, earlier realized revenue).
12. When presenting Gen to different levels of an organization, it's essential to tailor the message and provide concrete numbers and times for implementation.
13. Another challenge is navigating relationships within the organization, as colleagues may see Gen as encroaching on their areas of expertise.
14. Knowledge graphs in Gen can help address technology challenges by providing context and organizational knowledge, leading to more precise answers.
15. Graph databases are useful for managing complex relationships between data points and can be particularly helpful in industries like life sciences and manufacturing.
16. By combining vector and graph representations of data, users can get more contextually relevant results and improved explainability from their Gen applications.
17. Controlling access to information and understanding the relationships between different concepts are additional benefits of using knowledge graphs in Gen projects.

Source: AI Engineer via YouTube

❓ What do you think? What are your thoughts on the ideas shared in this video? Feel free to share your thoughts in the comments!