🌟 Editor's Note

Welcome to the AI Humble Servant Newsletter.

Google Talks to Dolphins: The Unexpected Direction of AI Research & OpenAI's Evolution

Editor's Note

Welcome to the AI Humble Servant Newsletter.

This week reveals two fascinating AI developments: Google translating dolphin sounds after 40 years of research, while OpenAI recalibrates how we access its models.

While Google builds bridges across species with DolphinGemma, OpenAI's strategic shift removes GPT-4.5 from its API (keeping it in ChatGPT) while launching GPT-4.1 with million-token context.

The message? AI expands in unexpected directions – both across species boundaries and within its ecosystems. For developers, the cost-effective GPT-4.1 becomes foundational. For researchers, GPT-4.5 remains in the curated interface.

This evolution presents us with choices about how we integrate these tools. Do we need raw API power or guided ChatGPT experience? Either way, these developments fundamentally transform what's possible.

🗓️ AHS Announcements!

OpenAI's Model Restructuring: Streamlining While Expanding

Editor's note: This major shift impacts AI workflows and strategies for businesses and creators in the coming months.

OpenAI's recent product announcements are creating renewed uncertainty in the AI industry with significant implications for businesses and creators.

• Model Retirement Timeline: OpenAI is phasing out GPT-4 from ChatGPT by April 30, and will remove GPT-4.5 from the API by July 14, 2025, though it will remain available in ChatGPT's research preview mode for paying users.

• Expanded Context Window: The new GPT-4.1 family can process up to 1 million tokens (approximately 750,000 words), enabling comprehensive analysis of entire document collections, codebases, or research papers in a single interaction.

• Performance Improvements: GPT-4.1 scores 54.6% on the SWE-bench Verified benchmark—a 21% improvement over GPT-4o and 27% over GPT-4.5, particularly excelling at coding tasks and instruction following.

• Cost Reductions: GPT-4.1 pricing is 26% lower than GPT-4o, while delivering comparable or better results than the more expensive GPT-4.5 ($75 per million input tokens, $150 per million output tokens).

• API-Only Accessibility: The new GPT-4.1 models are available exclusively through the API, not directly in the ChatGPT interface, signaling OpenAI's renewed focus on developer ecosystems rather than end-user applications.

AI Tool of the Week - Eleven Labs

thaddius_cho

678 followers

View more on Instagram

Transform Written Content into Lifelike Voice: Your Personal Audio Production Team

ElevenLabs has evolved from an AI voice experiment to an essential tool for content creators seeking to scale production without sacrificing authenticity, offering remarkably human-sounding voice generation across multiple languages.

The Backstory: Founded in 2022 by former Google and Palantir engineers, ElevenLabs has quickly risen to prominence in the AI audio space, securing $180 million in funding at a $3.3 billion valuation earlier this year. What began as a voice cloning project has expanded into a comprehensive audio production ecosystem.

Key Points:

• Lifelike Voice Generation: Creates natural-sounding audio with proper intonation, emotional nuance, and realistic pacing that can be difficult to distinguish from human narration

• Multi-Language Support: Offers voice generation across 32 languages, enabling global content distribution without hiring multiple voice actors

• Voice Cloning Capabilities: Allows creators to maintain voice consistency across all content by generating a custom voice based on short audio samples

• Conversational AI Integration: Recently launched voice agents that can handle interactive conversations for customer support, education, and entertainment applications

• Growing Media Partnerships: Adopted by major publishers including The New Yorker, The Washington Post, and The Atlantic for audio content production

Impact: Transforms content creation workflows by eliminating hours of recording and editing, allowing creators to focus on strategy and creativity while maintaining consistent voice quality across multiple projects.

User Access: Available through multiple subscription tiers starting at $5/month for the Starter plan (100,000 characters, approximately 2 hours of audio) up to Creator ($22/month) and Pro plans ($330/month) for professional content producers. A free tier provides limited access to test the technology.

Check out ElevenLabs

Real-World Application:

By handling the technical aspects of voice production, ElevenLabs allows creators to scale their audio content without sacrificing quality or authenticity—a perfect example of AI that augments human creativity rather than replacing it.

🔥 In Case You Missed It…

Sustainable Technology Developments

• AI's Energy Challenge: The International Energy Agency projects data centers will consume 945 terawatt-hours by 2030—equivalent to Japan's current annual electricity usage—with AI as the primary driver

• Green Transportation Alternative: Former SpaceX engineers at Parallel Systems are developing autonomous battery-powered rail vehicles that could eliminate up to 175 tons of CO2 per 100,000 miles compared to trucking

AI Ethics & Integration

• DeepMind's Future-Focused Hiring: Google DeepMind has posted a job opening for a Research Scientist to explore post-AGI societal implications, emphasizing the need for thoughtful planning around advanced AI

Human-Centered Technology

• Memory with Boundaries: OpenAI has introduced enhanced memory features for ChatGPT with thoughtful privacy controls, allowing users to disable memory, manage saved information, or use Temporary Chat for sensitive conversations

• Prompt Engineering Guidelines: Google has published a comprehensive whitepaper on effective prompting, emphasizing simple language, focusing on instructions rather than constraints, and controlling output formats

Surprising AI Applications

• Interspecies Communication: Google's DolphinGemma, built on 40 years of dolphin research, represents a significant step toward understanding non-human intelligence through AI

• Democratized Scientific Discovery: Seventeen-year-old Matteo Paz used AI to identify 1.5 million previously unknown space objects, demonstrating how AI tools are making scientific contributions possible for non-traditional researchers

LIVE AI TRAINING COMING SOON!

This will be a live training where we quickly go over how to start developing your AI voice!

Till next time,

AI Humble Servant

Thaddius Cho