Text to CAD AI is getting crazier

In partnership with

Google’s AI-Powered Daily Listen Creates Custom Podcasts

Google is rolling out a new “Daily Listen” feature that generates personalized, AI-powered podcasts based on your Discover feed. The feature is designed to keep you updated on topics you care about by creating short, customized audio episodes.

Currently available on the Google app for Android and iOS, Daily Listen is being tested with U.S. users who have joined Google’s Search Labs experiment.

What It Offers

• Short, Personalized Podcasts: Daily Listen creates podcast episodes up to five minutes long, summarizing key stories and topics tailored to your interests.

• Integrated Discover Experience: The topics you follow on Google’s Discover and Search platforms directly influence the content, ensuring relevance.

• Related Links: Alongside the audio, users can access clickable links to dive deeper into the topics covered.

• Real-Time Transcripts: Listen while reading a rolling written transcript, enhancing accessibility.

How It Works

If you have access to the feature, a new “Daily Listen” card will appear on the home screen of the Google app. Tapping the play button starts the podcast and shows the episode’s length. Users can scroll through related stories for additional context.

Inspired by NotebookLM

Google likens Daily Listen to its NotebookLM’s Audio Overviews feature, which generates podcasts with AI virtual hosts based on user-provided information like readings or legal briefs. However, Daily Listen focuses on delivering a quick, dynamic rundown of your personal interests.

This innovative feature aims to blend convenience and personalization, making it easier for users to stay informed through brief, curated audio experiences.

NVIDIA Unveils AI Blueprint for Retail Shopping Assistants

At NRF, NVIDIA announced the AI Blueprint for Retail Shopping Assistants, a generative AI workflow designed to transform the retail industry. Built on NVIDIA AI Enterprise and Omniverse™, the blueprint enables developers to create intelligent digital assistants that offer personalized recommendations and streamline shopping experiences.

Key Features

• Multimodal AI Capabilities: Supports text and image-based prompts for smarter searches.

• Realistic Product Previews: Uses Omniverse to visualize products in virtual spaces like your living room.

• Advanced Recommendations: Powered by NVIDIA NeMo™ microservices for accurate and contextual responses.

• Secure Interactions: Guardrails ensure safe and brand-aligned conversations.

Real-World Applications

• Helps browse product catalogs and quickly access detailed information.

• Features virtual try-ons for a more interactive shopping experience.

Availability

The AI Blueprint is available in early access, with developers encouraged to sign up for updates.

Zuckerberg Allegedly Approved Use of Pirated Books to Train Meta's AI Models

Authors, including Ta-Nehisi Coates and Sarah Silverman, have accused Mark Zuckerberg of endorsing the use of "pirated" books to train Meta's AI models in a US court filing. The allegations center on the use of the Library Genesis (LibGen) dataset, a "shadow library" containing millions of copyright-protected works.

Key Allegations

• Internal Approvals: A Meta memo cites Zuckerberg's approval of the LibGen dataset despite warnings from the company's AI executive team about its pirated nature.

• Regulatory Risks: Internal messages reveal concerns that using such datasets could weaken Meta’s negotiations with regulators.

• Data Access Hesitations: Engineers expressed discomfort about accessing LibGen data from Meta-owned laptops due to ethical and legal implications.

Legal Context

The authors suing Meta argue that the company misused their books to train Llama, its large language model. The lawsuit claims this infringes copyright laws and threatens the livelihoods of creative professionals.

A New York federal court previously ordered LibGen operators to pay $30 million for copyright violations, highlighting the contentious nature of using such datasets.

Ongoing Battle Over AI Training Data

The case reflects broader legal disputes over generative AI tools like ChatGPT and their use of copyrighted material. Creative professionals and publishers continue to challenge companies over the unauthorized use of their work in AI training.

A US district judge previously dismissed some claims against Meta but allowed the authors to amend their case, keeping the legal battle alive.

Text to CAD AI is getting crazier