This Week in Enterprise Tech (Audio) TWiET 561: That Cloud Looks Like A Llama - Moving on from old protocols, accessible machine learning with Predibase
Sep 16, 2023
Dev Rishi, Co-founder and CEO of Predibase, shares insights into making machine learning accessible for organizations. He tackles the challenges of moving large language models into production while emphasizing privacy concerns that hinder adoption. The conversation delves into the gap between interest and deployment of AI technologies, advocating for open-source solutions and cost-effective strategies. Rishi also discusses the importance of upgrading outdated protocols and how emerging innovations like AI Copilot are transforming security operations.
AI Snips
Chapters
Transcript
Episode notes
Old Tech Challenges Charities
- Older devices donated to charities often lack firmware upgrades and support for modern encryption.
- This creates security and usability challenges, forcing some organizations to discard donated equipment.
TLS Phasing Out Brings Complexity
- Phasing out TLS 1.0 and 1.1 improves security but causes compatibility issues.
- New TLS 1.3 features enhance security and performance, but removing older cryptographic algorithms raises migration challenges.
Privacy Blocks LLM Adoption
- Most organizations are highly interested in large language models (LLMs) but few have them in production yet.
- Privacy and ownership concerns significantly block the use of commercial LLMs in enterprises.