Daily AI Newsletter: Spotify experiments with an AI DJ that speaks Spanish

In today’s email:

🤫 $6M fine for robocaller who used AI to clone Biden’s voice
👀 Builder AI: The wild ride of a Microsoft-backed tech unicorn
🤑 Meta Is Working on a Paid Version of Its AI Assistant
🧰 11 new AI-powered tools and resources. Make sure to check the online version for the full list of tools.

Spotify experiments with an AI DJ that speaks Spanish

Spotify is developing a Spanish-speaking version of its AI DJ, currently referred to as "DJ Livi," aimed at expanding its language offerings beyond the original English-speaking AI DJ, DJ X. This development was uncovered by tech veteran Chris Messina, who found code snippets hinting at the new feature within the Spotify app. The code suggests that users could switch the AI DJ’s language from English to Spanish, and it’s noted that this feature might be available in Mexico if it rolls out.

Spotify has confirmed they conduct various tests to enhance user interaction through personalized music experiences, though they have not confirmed any specific plans regarding the launch of DJ Livi. The potential introduction of a Spanish-speaking AI DJ aligns with Spotify's strategy of utilizing AI for personalization, including experiments with AI in creating host-read ads and tailored playlists. The company continues to explore how AI can further influence its services in personalization, advertising, and content summaries.

Given the significant number of Spanish speakers in the U.S. and globally, where Spanish ranks as one of the top languages spoken, a Spanish AI DJ could cater well to a wide audience. While Spotify has yet to finalize the launch of this feature, the move could position Spotify strategically in key markets by enhancing user engagement through more personalized and linguistically diverse services.

NASA, IBM Research to Release New AI Model for Weather, Climate

NASA and IBM Research have jointly developed a new artificial intelligence model named the Prithvi-weather-climate foundational model. This innovative AI tool is designed to enhance various weather and climate applications by utilizing large, unlabeled datasets from NASA’s Modern-Era Retrospective analysis for Research and Applications (MERRA-2). The model leverages AI to improve the precision of regional and local weather and climate forecasting, making it possible to fine-tune it for diverse applications, including better management of severe weather patterns and natural disasters.

Sure, the collaboration extends across several notable institutions, including IBM Research, Oak Ridge National Laboratory, and various NASA departments. This open collaboration aims to democratize access to NASA’s extensive scientific data, enabling the global community to benefit from improved weather, seasonal, and climate projections. These efforts align with NASA’s mission to deliver actionable science that addresses the urgent changes observed on Earth, thereby aiding preparation, response, and mitigation strategies across the globe.

Sure, Prithvi-weather-climate is set to be part of a larger family of models, collectively known as the Prithvi family, which also includes models trained on harmonized data from NASA’s LandSat and Sentinel-2 missions. This model is expected to be released on the Hugging Face platform later this year, supporting NASA’s commitment to open science by making its data widely accessible and usable. This initiative not only fosters global collaboration but also enhances our ability to understand and predict changes in Earth’s climate and weather systems.

Introduction to On-Device AI - Short Course by DeepLearning

Enroll in the "Introduction to On-Device AI" course, offered in collaboration with Qualcomm, to learn how to deploy AI models directly on edge devices like smartphones and IoT devices. This beginner-friendly, one-hour course led by Krishna Sridhar, Senior Director of Engineering at Qualcomm, teaches you to utilize the local computing power of devices for faster, more secure AI processing. You'll learn to convert and quantize your PyTorch or TensorFlow models to enhance performance and reduce model size, ensuring compatibility with various devices.

The course covers essential topics such as reducing latency through on-device model deployment, preserving privacy, and optimizing efficiency. It introduces key concepts like neural network graph capture, on-device compilation, and hardware acceleration. Practical exercises include deploying a real-time image segmentation model on a device and integrating the model into an Android app, demonstrating the process step-by-step for effective learning.

Ideal for beginner AI developers, ML engineers, data scientists, and mobile developers with a basic understanding of Python and frameworks like PyTorch or TensorFlow, this course equips participants with the skills needed to deploy and optimize AI models on billions of edge devices efficiently. By completing this course, learners will be prepared to handle the increasing demand for on-device AI applications across various industries.

$6M fine for robocaller who used AI to clone Biden’s voice

The FCC has proposed a hefty $6 million fine against a scammer who impersonated President Biden using voice-cloning technology in robocalls during a New Hampshire primary election. These calls falsely instructed voters not to participate in the primary. The action underscores the agency's stance on the illegal use of AI in telecommunications to influence elections and deceive consumers.

The central figure in this operation, political consultant Steve Kramer, collaborated with the disreputable Life Corporation and various telecom aliases to execute these calls. Despite the significant proposed fine, Kramer and his associates have not yet faced criminal charges, reflecting the FCC's limited enforcement capabilities without the support of law enforcement agencies.

This case has prompted stricter regulations, with the FCC officially prohibiting the use of AI-generated voices in robocalls as of February. This decision came in the wake of growing concerns about the ease and potential consequences of generative AI technologies in creating convincing fake audio for fraudulent purposes.

Other stuff

Builder.ai: The wild ride of a Microsoft-backed tech unicorn
OpenAI Releases Former Staffers From Non-Disparagement Clauses
Microsoft’s Copilot AI search is making up fake Vladimir Putin quotes from press conferences that never happened
Meta Is Working on a Paid Version of Its AI Assistant
Microsoft outage affects Bing, Copilot, DuckDuckGo and ChatGPT internet search

Win 1 year free Superpower ChatGPT Pro🤑