Data Challenges and AI Impact in Research-Driven Organizations
summary
Claims that AI and modeling cut discovery time by integrating experiments with computation and automation, demonstrated by a 10–100x throughput high‑throughput screening setup that couples DFT, mesoscale and finite‑element models on HPC to span molecular to plant scale. A production workflow and metadata schema still miss instrument, method and calibration details, creating annotation gaps. Governance balances access vs security, tool proliferation creates tech debt, and a hiring pipeline brings 50–60 dual‑domain interns yearly to sustain a DFT-to-plant-scale HPC modeling stack.
AI-written summary · about
chapters (16)
- 0:00Introduction and overview
- 3:23Digital transformation at Dow
- 5:07Evolution of experimental methods
- 9:00Role of modeling and AI
- 11:50Data challenges in research
- 15:59High-throughput data management
- 17:01Data access and security balance
- 18:48Tool standardization challenges
- 19:48Upskilling the workforce
- 23:19Integrated tools and governance
- 24:44Partnerships and future outlook
- 25:38Q&A on data access and generative AI
- 27:00Q&A on change management and training
- 31:31Q&A on business value connection
- 34:19Q&A on semantics and LLMs
- 36:10Q&A on autonomous experimentation
related talks
30mAI in Drug Discovery
28mData Curation: Transforming Bytes into AI Gold
16mOperationalizing AI in Finance: Real-Time Analytics
08mLeveraging data insights to increase efficiency and innovation
31mAI, Digital and Value Streams Enable Data Driven Organizations
46mThe AI dominated world – skills required for successfully navigating it
11mData-centric AI: A new paradigm
28mAI/ML for Non Data Scientists