Daily AI Newsletter: Stability AI, Facing Cash Crunch, Discusses Sale

In today’s email:

🧠 How does ChatGPT ‘think’? Psychology and neuroscience crack open AI large language models
When AI helps you code, who owns the finished product?
🔥 Chrome Dev Tools: Understand errors and warnings better with Gemini
🧰 8 new AI-powered tools and resources. Make sure to check the online version for the full list of tools.

Instagram’s co-founder is Anthropic’s new chief product officer

Mike Krieger, co-founder of Instagram and former head of the AI news-reading app Artifact, has been appointed as the new chief product officer at Anthropic. Announced today, Krieger will oversee all of Anthropic’s product initiatives, marking a pivotal moment for the company. Anthropic, founded by ex-OpenAI employees, has been primarily focused on developing its core technology. Now, with the recent release of the Claude app for iOS and multilingual support, the company is shifting its focus to product development to keep pace with competitors like OpenAI, Google, Microsoft, and Apple.

Anthropic CEO Dario Amodei highlighted Krieger’s expertise in creating user-friendly products, emphasizing the potential for developing new ways for users to interact with Claude, particularly in professional settings. This suggests that Krieger’s initial focus may be on finding business-oriented applications for Claude. Despite being somewhat slower to market, Anthropic is well-positioned with nearly $8 billion in funding, substantial investments from Amazon and Google, and potential collaboration with Apple.

The AI industry is currently experiencing rapid growth and intense competition, with new advancements and significant hype emerging daily. Krieger, who previously built a successful company in a highly competitive environment, now faces even higher stakes at Anthropic. His experience and vision are expected to drive the company forward as it navigates this dynamic and fast-evolving landscape.

Google CEO Pichai says company will ‘sort it out’ if OpenAI misused YouTube for AI training

Google CEO Sundar Pichai addressed concerns about OpenAI potentially using YouTube content to train its AI models. During an interview with CNBC's Deirdre Bosa, Pichai stated that Google would "sort it out" if it found that OpenAI, backed by Microsoft, had violated YouTube’s terms of service. This follows comments from OpenAI's technology chief, Mira Murati, who was unsure if YouTube videos were included in the training data for their Sora model. Pichai emphasized that Google has clear terms of service and processes to ensure compliance, though he did not specify if legal action would be pursued.

This conversation occurred alongside the announcement of Google’s new AI model, Veo, at its I/O developer conference. Veo can generate synthetic videos, and early access will be granted by approval. The timing of this announcement closely followed OpenAI's unveiling of its GPT-4o model, which introduces advanced features such as realistic voice conversations and image analysis via a smartphone camera. Pichai pointed out that OpenAI’s new features have not yet been released to users, while Google's similar capabilities for its Gemini chatbot will be available later this year.

Pichai also highlighted Google’s advancements in AI, including an 80% reduction in the cost of serving AI models in web searches, thanks to custom TPUs and Nvidia GPUs. He mentioned that Google’s AI Overviews in search results are now available to all U.S. users. Addressing future collaborations, Pichai noted Google's ongoing partnership with Apple and the potential integration of their AI technologies into Apple's ecosystem, aiming to enhance user experiences across platforms.

Stability AI, Facing Cash Crunch, Discusses Sale

British artificial intelligence startup Stability AI has been in discussions with at least one potential buyer in recent weeks amid a significant cash crunch, according to a report by The Information. In the first quarter of 2024, the company generated less than $5 million in revenue while incurring losses exceeding $30 million. Additionally, Stability AI currently owes close to $100 million to cloud computing providers and other creditors.

The company's financial struggles have led to a search for potential buyers and a restructuring process that included layoffs. This follows the recent resignation of founder Emad Mostaque as the company's CEO. Stability AI, known for releasing AI models that generate audio and video from text prompts, has been facing mounting financial pressures despite its technological advancements.

Stability AI, which employs nearly 200 people globally, did not immediately respond to Reuters' request for comment on the report. The company’s ongoing financial challenges and restructuring efforts highlight the difficulties faced by AI startups in balancing rapid innovation with financial stability. The situation underscores the broader challenges within the AI industry as companies strive to manage costs while pushing the boundaries of technology.

Chrome Dev Tools: Understand errors and warnings better with Gemini

To enhance your understanding of errors and warnings in the Console, Chrome DevTools offers detailed explanations via an AI feature. Ensure you are using Chrome version 125 or later, signed in with your Google Account, and have English (US) selected in settings. Enable the 'Understand console messages with AI' option in DevTools settings under Preferences > Console. If requirements aren't met, an info icon will provide details.

To use this feature, visit a demo page, open DevTools, and inspect the console messages. Hover over a console error to see the 'Understand this error' button. Click it to open a dialog that explains the error, following the privacy and legal notices. If explanations are unsatisfactory, use the search option for further details. Provide feedback to help improve the feature.

Chrome DevTools uses your console message data, stack trace, source code, and network headers to generate explanations. Google collects and processes this data to improve its products and services. The feature is experimental and may produce inaccurate results. Always double-check the generated explanations and use code snippets cautiously. The feature is subject to future changes and improvements.

Other stuff

How does ChatGPT ‘think’? Psychology and neuroscience crack open AI large language models
When AI helps you code, who owns the finished product?
We have to stop ignoring AI’s hallucination problem
Easter Eggs of AI. Meme References, Duplicates, Biases, and Other AI Hallucinations and Why They Happen
Lang Chiain fundamentals

Win 1 year free Superpower ChatGPT Pro🤑