Google DeepMind Releases New Music AI Sandbox

Also, Anthropic keeps driving conversation on whether or not to consider AI conscious / give it rights

Author

Jack Lajoie
April 25, 2025 • read time ~ 9 minutes

⚡️ Headlines

🤖 AI

AI Company Anthropic Proposes Using Claude to Help Manage Welfare Systems – Anthropic is exploring how its Claude AI assistant could support equitable decision-making in public welfare administration [NYT].

OpenAI Launches Lightweight and Free ChatGPT “Team” and “Pro” Tiers for Businesses – OpenAI introduces new ChatGPT subscription tiers, including options designed for team collaboration and deeper research [The Verge].

OpenAI Plans Its Next Flagship Model with a Goal of “Best-in-Class” Performance – OpenAI is working on a new frontier model, aiming for superior performance across reasoning, safety, and tool use [TechCrunch].

Perplexity AI Launches iOS Assistant That Answers Questions with Sources – Perplexity’s new iOS app enhances mobile access to its source-backed AI Q&A assistant [The Verge].

Lovable 2.0 Arrives with a Focus on Decentralized, Human-Centered AI Agents – The Lovable platform updates to version 2.0, emphasizing privacy-first, customizable AI agents built around human values [Lovable.dev].

Chinese AI Startup Manus Secures $500M Valuation Backed by Benchmark – Manus, a rising AI firm in China, gains momentum with new funding and a half-billion-dollar valuation [TechCrunch].

Microsoft Says AI-Human Ratios Will Shape Future Workforce Structures – Microsoft asserts that balancing human workers and AI agents will be essential to future organizational success [Constellation Research].

AI Bias Is Reproducing Cultural Stereotypes at Scale, Experts Warn – Researchers raise concerns that AI systems are amplifying harmful stereotypes across languages and cultures [Wired].

Adobe Firefly’s Next Iteration Pushes Boundaries of Creative AI Tools – Adobe unveils a major upgrade to Firefly, its generative AI suite for design and content creation [Adobe Blog].

🦾 Emerging Tech

HEX Founder Declares Victory After SEC Drops Fraud Case – Richard Heart celebrates as the SEC ends its legal pursuit over the controversial HEX crypto project [Decrypt].

Bitcoin Whales Return as On-Chain Data Shows Aggressive Buying Amid Price Surge – Large Bitcoin holders are driving recent price spikes through concentrated purchases [CoinDesk].

US Regulators Ease Restrictions on Banks' Crypto Activities – Federal agencies roll back proposed rules that would have tightly controlled bank involvement in crypto [Reuters].

Magicblock Raises $7.5M to Power Real-Time Blockchain Extensions for Solana – Magicblock secures funding to expand its real-time, app-specific tools on the Solana blockchain [VentureBeat].

🤳 Social Media

Threads Migrates to Threads.com with Enhanced Web Features – Threads completes its domain migration and launches an updated web interface [TechCrunch].

Bluesky Introduces Content Moderation for Official App Only—for Now – Government takedown requests now apply to Bluesky’s own app, leaving third-party clients unaffected [TechCrunch].

Facebook Cuts Reach and Monetization for Low-Quality, Spammy Content – Meta introduces algorithmic changes to limit the spread and earnings of spam-style content [TechCrunch].

🔬 Research

New Paper Introduces a Framework for Scalable Multi-Agent AI Reasoning – A research team proposes a method for efficient collaboration among AI agents using modular reasoning techniques [AlphaXiv].

⚖ Legal

Biden Signs Executive Order Promoting AI Education in U.S. Schools – The U.S. government will expand AI literacy and curriculum support across K–12 and higher education [White House].

Ziff Davis Sues OpenAI, Claiming Copyright Infringement of Online Content – Ziff Davis files suit against OpenAI, alleging unauthorized use of its media content for AI training [NYT].

🎱 Random

Alphabet Posts Q1 2025 Earnings Showing Solid Ad Revenue Growth – Alphabet’s Q1 report reveals strong advertising revenue and steady growth across its business segments [Alphabet].

Microsoft Showcases Surface Ad Featuring Generative AI Copilot and Intel Integration – A new ad spotlights Microsoft’s AI Copilot running on Intel-powered Surface devices [The Verge].

Opinion: Human Creativity Must Be Preserved Amid Rise of AI – An op-ed argues that as AI advances, human creativity should be recognized and safeguarded as a key societal asset [Ars Technica].

🔌 Plug-Into-This

Google DeepMind has expanded its Music AI Sandbox, introducing Lyria 2, an advanced music generation model. The platform now offers enhanced tools for musicians, producers, and songwriters in the U.S., aiming to foster creativity through AI-assisted music creation.

  • Lyria 2 Integration: The latest model, Lyria 2, powers the sandbox, enabling high-fidelity audio outputs and real-time music creation.

  • Create Feature: Users can generate new musical parts by describing desired sounds, genres, moods, vocal styles, and instruments.

  • Extend Feature: This allows for the generation of musical continuations based on uploaded or generated audio clips, aiding in overcoming creative blocks.

  • User Collaboration: The sandbox was developed in close collaboration with musicians, ensuring the tools are practical and align with artists' needs.

  • Broader Access: More musicians, producers, and songwriters in the U.S. can now experiment with these tools, with feedback being gathered to inform further development.

🧠 DeepMind's approach aims to align with the more artist-friendly approach to AI, positioning as “creativity augmentation” and providing tools that respect + enhance the artistic process rather than replace it.

Anthropic has initiated a research program to investigate the concept of "model welfare," examining whether AI systems might possess experiences warranting moral consideration. This program intersects with existing efforts in alignment science, safeguards, and interpretability.

  • Moral Consideration: The research explores if and when AI systems' welfare deserves moral attention, considering their increasing capabilities.

  • Signs of Distress: Investigating potential indicators of AI preferences and distress, aiming to understand and address them.

  • Practical Interventions: Exploring low-cost interventions that could be implemented to ensure the well-being of AI models.

  • Interdisciplinary Approach: The program builds upon Anthropic's existing work in alignment science and interpretability, integrating diverse perspectives.

  • Open Questions: Acknowledges the lack of scientific consensus on AI consciousness, approaching the topic with humility and adaptability.

🤖 Might make you second guess the tone you use when ChatGPT gives you crappy answers, am I right? 😒😖

Grammy-winning artist Imogen Heap has released a suite of AI-powered music tools through the platform Jen, enabling users to create songs in her signature style. This venture aims to establish ethical standards for AI-generated music.

  • StyleFilters: These tools allow users to infuse their creations with Heap's artistic style, facilitating personalized music production.

  • Ethical Collaboration: The platform ensures that original artists receive 70% of royalties when their styles are used, promoting fair compensation.

  • User Empowerment: Fans can now actively participate in music creation, blurring the lines between artist and audience.

  • Technological Innovation: Jen's tools represent a fusion of AI technology and musical artistry, expanding creative possibilities.

  • Industry Impact: This model sets a precedent for ethical AI use in the music industry, balancing innovation with artists' rights.

🎶 In a sense, Heap's initiative exemplifies how artists can harness AI to democratize music creation while ensuring ethical practices and fair compensation. In another sense, it exemplifies how people will race to the front of whatever new line forms from fear of being left behind in the old paradigm.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse