The AI Whisperer, Poaching AI Talent, Improving CX with AI, Strengthening AI System Security

This Week in AI: Volume CI, Issue #27

Meet the AI Expert Advising the White House, JPMorgan, Google and the Rest of Corporate America

Ethan Mollick is sociologist on the wrong coast. He has become the AI whisperer to those most responsible for building, using, and defining the future of AI.

• Mollick’s background is not in computer science or AI, but in sociology and economics; however, his unique perspective and his ability to see the bigger picture have made him a resource for both companies and policymakers alike.

Mollick views the future of AI with optimism and excitement. He believes AI will revolutionize many aspects of our lives and society, so we should engage deeply with the technology to shape its development. Now is the time to look at the potential upside of AI and shape its future direction.

Apple poaches AI experts from Google, creates secretive European AI lab

The potential technological advancement driving Apple’s recruitment strategy is generative AI.

• Over the past couple of years Apple has been poaching dozens of artificial intelligence experts from Google while building a secretive lab in Zurich to develop new AI models and products.
• Zurich-based employees have been involved in Apple’s research into the underlying technology that powers AI products such as OpenAI’s ChatGPT.
• Apple has acquired two local AI startups in Zurich, FaceShift and Fashwell, to build its research laboratory in the city.
• Industry insiders say Apple is focused on deploying generative AI on its mobile devices; this breakthrough would allow AI chatbots and apps to run on the phone’s own hardware and software rather than be powered by cloud services in data centers.
• Apple’s top AI team is made up of former key figures from Google, including Giannandrea, who previously oversaw Google Brain, the search company’s AI lab which subsequently merged with DeepMind.

NSA Publishes Guidance for Strengthening AI System Security

NSA is emphasizing the importance of protecting model weights. Model weights are the learnable parameters of a deep neural network. They represent the results costly investments in training advanced AI models. This includes significant compute resources; collected, processed, and potentially sensitive training data; and algorithmic optimizations. NSA recommends organizations limit and protect access to model weights, as they are the essence of the AI system even though most organizations focus on protecting data and algorithms. Model weights are just as important.

To mitigate model compromise and other attacks against model weights, tech companies need to:

• Harden the interfaces for accessing model weights and make it more difficult for attackers to exfiltrate them.
• Implement hardware protections for model weight storage, protecting against emanation or side channel techniques.
• Aggressively isolate weight storage to prevent unauthorized access.

COOL TOOLS

Intellecta AI

Improve Intellecta AI instantly responds to customers via chat, social media, or email — in any language:

GeoSpy AI

This is insane. Upload a picture GeoSpy AI and it’s freaky accurate with geolocating. Not perfect of course, and my unscientific research shows it’s pretty dang accurate; at least to the State level. It looks like the AI is processing the environment to include reading signs. Spooky stuff.