Google’s Whisk AI tool enables photo-to-prompt creativity, OpenAI introduces o1 reasoning model with advanced tools for developers

⚡️ Quick Hits

🤖 AI

NVIDIA Introduces Jetson Orin Nano Super: Affordable Generative AI Supercomputer - NVIDIA's new Jetson Orin Nano Super Developer Kit offers enhanced generative AI capabilities at a reduced price of $249, delivering up to 1.7x performance improvements for developers and hobbyists. [NVIDIA Blog].

Google's Code Assist Expands with Third-Party Tool Integrations - Google's enterprise-focused code assistant, Code Assist, now supports third-party tools, enhancing its utility for developers by integrating with popular development environments. [TechCrunch].

Talkie Eyes AI Video Features Amid Growth Challenges - Talkie, a Chinese-developed AI app rivaling Character.AI, plans to introduce AI video capabilities to reignite growth after a slowdown in monthly active users. [The Information].

🎨 Creative

YouTube to Detect and Manage AI-Generated Content of Creators - YouTube, in collaboration with Creative Artists Agency (CAA), plans to implement technology to identify and manage AI-generated content featuring celebrities and creators, aiming to protect their digital likenesses. [Variety].

₿ Crypto

Bitcoin Surges Past $108,000 Amid Institutional Interest - Bitcoin's price has exceeded $108,000, with analysts suggesting that institutional investments could drive it towards $210,000 in the near future. [Crypto.News].

Arthur Hayes Predicts Crypto Market Correction Post-Trump Inauguration - BitMEX co-founder Arthur Hayes forecasts a significant cryptocurrency market downturn following President-elect Donald Trump's inauguration, advising investors to prepare for potential volatility. [Cointelegraph].

⚖ Legal

OpenAI's Transition to For-Profit Model Raises Governance Concerns - OpenAI's shift from a nonprofit to a for-profit entity has sparked debates over its governance structure and commitment to its original mission of advancing AI for the benefit of humanity. [The New York Times].

Trump Administration Plans Overhaul of H-1B Visa Program - The incoming Trump administration is set to implement significant changes to the H-1B visa program, potentially impacting the tech industry's ability to hire foreign talent. [The New York Times].

🧪 Research

New Study Evaluates Stability of Reasoning in Large Language Models - Researchers have introduced 'G-Pass@k,' a novel metric to assess the reasoning stability of large language models, highlighting the need for more robust evaluation methods in AI development. [arXiv].

🎱 Random

Apple Maps Web App Adds Look Around Feature - Apple has updated its Maps web application to include the Look Around feature, offering users interactive, street-level imagery directly from their browsers. [9to5Mac].

You can now call ChatGPT at 1-800-ChatGPT - or by sending a WhatsApp to the same number. [X]

🔌 Plug In To These Details

Google’s Whisk AI Tool Enables Image-Based Prompts

Google’s new AI tool, Whisk, allows users to generate images by using existing photos as creative prompts. The system integrates advanced AI models to provide quick and intuitive visual exploration options for designers and casual users alike.

• Whisk lets users upload photos to define subject, style, or background, using multiple images to refine outcomes.

• Gemini AI translates uploaded images into text-based descriptions, which Imagen 3 then uses to create new images.

• Users can incorporate text prompts alongside photos for more precise image generation.

• The tool offers AI-suggested photos for users who don’t have suitable starting images.

• Refinement options enable users to tweak prompts and regenerate results iteratively.

🎨 Whisk makes creative experimentation with AI images even more accessible and fast, bridging the gap between vision and creation for a wider range of users.

OpenAI Introduces o1 and Developer Tools Enhancements

OpenAI has unveiled o1, a reasoning model designed to handle complex, multi-step tasks with advanced accuracy. Accompanying this release are several developer tools aimed at enhancing performance, flexibility, and cost-efficiency in AI-driven applications.

• o1 Model Features: Supports function calling, developer messages, structured outputs, and vision capabilities, enabling seamless integration with external data and APIs.

• Realtime API Updates: Introduces WebRTC support for low-latency, real-time voice interactions, a 60% price reduction for GPT-4o audio, and support for GPT-4o mini at reduced rates.

• Preference Fine-Tuning: A new customization technique allowing developers to tailor models based on specific user and developer preferences.

• New SDKs: Beta releases of Go and Java SDKs to facilitate integration into diverse development environments.

🚀 These advancements aim to streamline AI application development, offering developers enhanced tools for building sophisticated, real-time conversational experiences.

Databricks Achieves $62 Billion Valuation Under CEO Ali Ghodsi

Databricks, led by CEO Ali Ghodsi, has reached a valuation of $62 billion, evolving from offering free software to a robust enterprise platform utilized by major corporations.

• Strategic Shift: In 2016, Ghodsi transitioned Databricks from free software to a paid model, enhancing features to attract large clients like Walgreens and Rivian.

• Microsoft Partnership: A pivotal 2017 deal with Microsoft integrated Databricks into Azure, significantly boosting sales and market reach.

• Operational Efficiency: Ghodsi implemented measures such as offshoring jobs and developing AI bots to streamline operations and improve productivity.

• Acquisitions for Growth: Recent acquisitions, including AI startup MosaicML for $1.3 billion, have expanded Databricks' capabilities in AI and data management.

• Funding Decisions: Opting for a $10 billion private funding round over an IPO, Ghodsi aims to support company growth and provide employee incentives.

💸 Databricks' record-breaking $10 billion VC round—the largest in history—is another indicator of the continued intense momentum in AI. Surely a bubble sign as well, right?

📸 Creator Corner

Midjourney Moodboards

Midjourney, the renowned AI image generator, has recently introduced new features that significantly enhance creative workflows:

Pinterest-Inspired Moodboards

Users can now upload curated image collections as "moodboards," serving as inspiration for generating new art. The AI adapts to the visual elements of these images, creating unique style profiles that reflect the user's aesthetic preferences.

Multiple Personalization Profiles

The platform now supports multiple personalization profiles, allowing users to create and switch between custom versions of Midjourney's latest AI model, version 6.1. This facilitates seamless integration of personalized styles across various projects.

Streamlined Custom Model Setup

Setting up a custom model has become more efficient, with a fivefold improvement in image ranking speed. Users need just 40 image ratings to begin creating a profile, achieving optimal stability at 200 ratings. This streamlined process lowers the barrier to entry for new users, enabling quicker personalization.

🤔 Final Thoughts

With record-breaking valuations and more dev focused releases, the question remains: are we building sustainable systems, or inflating a bubble primed for a painful correction?

The speed at which these products seem to improve could legitimately put the industry in a position to outrun the physics that have governed tech thus far.

~ JL