Make AI Work More Than 5% of the Time
Ihre Präsentator(en)
Lexi Reese
CEO & Co-founder at Lanai
Lexi is building Lanai, an empowerment platform for human-Gen AI collaboration. She has two decades experience in technology leadership. Lexi ran for United States Senate Candidate for the State of California in 2023-24. Previously, she was Executive In Residence at General Catalyse, Chief Operating Officer at Gusto, and VP of Global Programmatic Platforms at Google.
Pratyush Mulukutla
CDO at MediaMint
Pratyush runs the data department at MediaMint, after his data and AI consultancy, DataBeat, was acquired earlier this year. He has two decades of experience as a consulting leader, with a focus on building AI, agentic systems, advanced analytics, and reporting solutions. Previously, Pratyush was Chief of Staff at Deloitte India, and he is a recipient of the prestigious 40 Under 40 Indian Achievers Club award.
Jon Welch
Vice President of AI & ML at Albert Invent
Jon runs the machine learning and AI team at materials science company Albert Invent. His work involves constructing large scale cloud data science architectures and generative modeling. Previously, Jon was Director of Vision Technologies at edge voice control company Sensory and Principal Data Scientist at identity management platform TruU.
Session Resources
Summary
The session "Make AI Work More Than 5% of the Time" focuses on improving AI project success rates for data professionals and business leaders. It addresses the concerning statistic from an MIT study that claims 95% of AI pilots fail to reach production. The discussion features insights from Lexi Reese, CEO of Lanai, Pratyush Mulukutla, Chief Data Officer at MediaMint, and Jon Welch, VP of AI at Albert Invent. They explore common pitfalls in AI projects, share success stories, and provide guidance on building the right infrastructure and skills to enhance AI adoption. The session emphasizes the importance of aligning AI initiatives with clear business objectives and measuring their impact effectively.
Key Takeaways:
- AI success requires clear business objectives and measurable outcomes.
- Understanding the limitations and scope of AI is essential for effective implementation.
- Building a strong data infrastructure is foundational to AI success.
- Translating data insights into business value is a critical skill.
- AI projects should start with practical, achievable goals and user-centric design.
Deep Insights
Understanding AI Project Failures
The session begins by addressing the startling statistic from MIT that 95% of AI pilots fail to reach production. Lexi Reese dismisses the study as overly pessimistic, suggesting that many AI projects do provide value, but their impact is often unmeasured. Pratyush Mulukutla agrees, noting that AI's success is often hindered by unclear objectives and inadequate data infrastructure. Jon Welch adds that unrealistic expectations can lead to perceived failures, as companies often set overly ambitious goals without understanding the incremental nature of AI improvements. The consensus is that AI projects need clear business objectives and realistic expectations to succeed.
Common Mistakes in AI Implementation
A recurring theme is the importance of aligning AI projects with business goals. Pratyush highlights the mistake of pursuing AI for its own sake, without a clear business problem to solve. This often leads to pilots that never progress to production. Instead, companies should identify specific workflows where AI can provide measurable efficiency gains. Lexi emphasizes the need for strategic leadership in AI adoption, suggesting that businesses should integrate AI into their core strategy rather than treating it as a side project. Jon stresses the importance of understanding the limitations of AI and setting achievable goals.
Success Stories and Practical Applications
The speakers share various success stories to illustrate effective AI use. Jon discusses a project with Nureon, where AI was used to enhance search capabilities for raw materials, demonstrating the importance of domain-specific AI applications. Lexi mentions T-Mobile's 30% efficiency gain across business functions after building a unified AI data factory. These examples highlight the need for specific AI solutions that address business needs. The speakers agree that successful AI projects often start with practical, achievable goals and involve close collaboration with end-users to ensure the solutions meet their needs.
Building the Right Infrastructure and Skills
The discussion emphasizes the necessity of a strong data infrastructure as a foundation for AI success. Lexi points out that every successful AI story begins with a solid data architecture. Pratyush adds that having a clear understanding of the data environment and the right technical skills, such as data engineering and MLOps, are essential. The speakers also highlight the importance of translating data insights into business value, a skill that is in high demand. They stress that AI initiatives should be driven by business needs and supported by a comprehensive understanding of both the technology and the business context.