How AI helped train Amazon’s palm scan payment technology

Welcome to Daily Zaps — It’s Wednesday already. We love short weeks! 🙌

Here’s what we got for you today:

How AI helped train Amazon’s palm scan payment technology

Amazon One is being introduced in over 500 Whole Foods Market stores and other locations, recognizing unique palm features using infrared light and vein patterns to create a secure palm signature for identification.

To tackle the challenge of accuracy in palm recognition, Amazon turned to generative AI and synthetic data. By generating millions of synthetic palm images, the AI model was trained to discern varying conditions and annotations, including rejecting fake hands and increasing system accuracy. [video explanation here]

Amazon One's applications extend beyond payments, encompassing loyalty programs and age verification. The innovation's privacy-centric design ensures it operates outside normal light spectrum, respecting customer privacy by not using palm information for identification.

Together With Our Partners

Tech news your CEO reads

The Information is where you go to get the most important news about technology and business first. It’s what leaders read to stay one step ahead, and to gain market-moving insight into what’s next, of consequence and least understood—before everyone else does.

Subscribe today and save up to $250.

This tropical island is cashing in on dot ai domain names

The Caribbean island of Anguilla, famous for its natural beauty, has unexpectedly become a digital hotspot due to its control over internet addresses ending in .ai. Previously overlooked, this domain has surged in popularity among artificial intelligence (AI) startups and tech giants alike. Notable names like Stability.ai, Elon Musk's X.ai, and Character.ai have secured .ai domains. The total registrations for .ai domains have doubled to 287,432 in the past year, generating an estimated $30 million in domain registration fees for Anguilla in 2023.

The island's .ai domains gained prominence after the rise of artificial intelligence in various industries, evident from the demand for catchy and brand-relevant web addresses. The island's economy, reliant on tourism, experienced a boost in revenue due to the increased demand for .ai domains and the planned rise in registration fees.

Everybody will have their own AI-powered personal assistant

Mustafa Suleyman, co-founder of DeepMind and current CEO of Inflection AI, predicts that within the next five years, everyone will have their own AI-powered personal assistants. He believes that as AI technology becomes more affordable and widespread, people will have access to smart AI assistants that "know" them, understand their personal history, and help manage their daily lives. Suleyman envisions AI assistants that can reason, prioritize tasks, aid in creativity, and serve as companions.

He likens these AI assistants to "chiefs of staff" that intimately understand individuals' interests, manage their information needs, and assist with decision-making. Suleyman believes that AI's role will extend beyond personal assistance, impacting various aspects of people's lives, much like how the internet transformed society. This perspective aligns with the views of other tech leaders like Bill Gates and Tim Cook, who recognize the revolutionary potential of AI to simplify tasks and enhance productivity across different domains.

AI can predicts a chemical’s smell

An artificial intelligence (AI) system has been created to describe smells based on their molecular structures, often producing similar descriptions to those given by trained human smell testers. The system, developed by Osmo, a startup derived from Google Research, uses a neural network to assign words like 'fruity' or 'grassy' to around 5,000 odorants according to their chemical makeup. The AI's descriptions closely matched human responses, suggesting it could help design new scents and offer insights into how the human brain perceives smells.

The AI identified connections between specific chemical patterns and certain smells, forming a principal odour map (POM). This POM was tested against human data, where 15 volunteers matched words to smells. The AI's predictions aligned well with human responses, indicating potential uses in industries like food and cleaning products.