Why Traditional AI Fails in Drug Discovery - Elucidata's Data-Centric AI Approach

AI is great at pattern matching — but what happens when the most valuable insights don’t fit the pattern?

In this episode of DataStorage.com, we sit down with Abhishek "AJ" Jha "(AJ), Founder & CEO of Elucidata, to break down why traditional AI approaches fail in drug discovery — and how data-centric AI is reshaping the future of pharma, healthcare, and beyond.

AJ shares Elucidata’s journey from resisting the “AI” label to fully embracing it in the post-LLM era — not by building bigger models, but by focusing on data quality, governance, and out-of-distribution problems that actually matter in regulated industries.

🔍 In this conversation, we cover:
 • Why pattern-matching AI breaks down in drug discovery and healthcare
 • What “out-of-distribution” problems are — and why they’re so valuable
 • How data-centric AI differs from model-centric AI
 • The role of human-in-the-loop AI in high-stakes industries
 • Preparing multimodal data (text, tabular, imaging) for AI-ready use cases
 • Why starting with the use case beats “cleaning all the data”
 • AI infrastructure decisions, cloud costs, GPUs, and egress challenges
 • Deploying AI securely in regulated environments (HIPAA, SOC 2, GDPR)
 • Whether companies should build their own models or fine-tune existing ones

This episode is a must-watch for:
 • Pharma & biotech leaders
 • AI and data infrastructure teams
 • Founders building AI products in regulated industries
 • Anyone questioning whether bigger models = better AI

⸻

🎧 Subscribe to DataStorage.com for deep conversations on AI, cloud infrastructure, data storage, and the real systems powering the AI economy.

🔗 Learn more about Elucidata: https://elucidata.io
🔗 More episodes: https://datastorage.com