Dexa/Demetrios Brinkmann

Published Feb 7, 2025

Robustness, Detectability, and Data Privacy in AI // Vinu Sankar Sadasivan // #289

Vinu Sankar Sadasivan, a student researcher at Google DeepMind, delves into the intricacies of AI text detection, model vulnerabilities, and defense mechanisms, highlighting the challenges in developing reliable methodologies amidst evolving AI capabilities and attacks. This episode unveils the advanced techniques employed to enhance AI robustness, detectability, and data privacy, emphasizing the critical need for ongoing security improvements.

Episode Highlights

Topics covered

Questions from this episode

- What are some applications of language models?
  Asked by 7 people
- What has been said about artificial intelligence?
  Asked by 3 people
- What techniques are used with large language models (LLMs)?
  Asked by 2 people
- What is the best insight on prompt engineering and engaging large language models (LLMs) from the episode John Maeda: AI + Design and the clip Prompt Engineering Insights?
  Asked by 1 person
- What is the latest trend in AI?
  Asked by 1 person
- Can we detect hostile motivations in AI?
  Asked by 1 person
- How can we mitigate AI harm?
  Asked by 1 person
- How can AI manipulate humans?
  Asked by 1 person
- How could AI transform communication as discussed in the episode Jacob Howland on the Hidden Human Costs of AI and the clip The Future of Writing?
  Asked by 1 person
- How could AI be subverted?
  Asked by 1 person
- Tell me something unique about large language models
  Asked by 1 person
- How reliable is AI-generated content?
  Asked by 1 person
- What is the best insight on prompt engineering and engaging large language models (LLMs)?
  Asked by 1 person
- What techniques are used with large language models (LLMs) in the episode Dodging Latent Space Detectors: Obfuscated Activation Attacks with Luke Bailey, Erik Jenner, an... and the clip Understanding Model Activations?
  Asked by 1 person
- What are the dangers of language models?
  Asked by 1 person

Episode Highlights

Related Episodes

Retrieval Augmented Generation
Answers 383 questions
When Machine Learning meets privacy - Episode 4
Answers 383 questions
When Machine Learning meets Data Privacy
Answers 383 questions
AI For Good - Detecting Harmful Content at Scale // Matar Haller // #245
Answers 383 questions
Extending AI: From Industry to Innovation // Sophia Rowland & David Weik // #246
Answers 383 questions
AI Agents for Consumers // Shaun Wei // #244
Answers 383 questions
AI in Healthcare // Eric Landry // #249
Answers 383 questions
Security and Privacy // Day 2 Panel 1 // AI in Production Conference
Answers 383 questions
Data Quality = Quality AI // AIQCON Panel
Answers 383 questions
Machine Learning, AI Agents, and Autonomy // Egor Kraev // #282
Answers 383 questions
Harnessing AI APIs for Safer, Accurate, & Reliable Applications // Ron Heichman // #252
Answers 383 questions
AI Operations Without Fundamental Engineering Discipline // Nikhil Suresh // #250
Answers 383 questions
When Machine Learning meets privacy - Episode 3 with Charles Radclyffe
Answers 383 questions
Data Governance and AI // Alexandra Diem // #212
Answers 383 questions
When Machine Learning meets privacy - Episode 5
Answers 383 questions