Our 138th episode with John Kron discussing last week's AI news, including DALLE-3's image generation, Gemini surpassing Chat GPT-4, China's semiconductor breakthrough, SoftBank's potential investment in OpenAI, challenges of investing in hardware companies, Databricks' fundraising and Lakehouse IQ, Meta's plan for next-gen model, race dynamics and multimodal capabilities in AI, AI in scientific research, limitations of context windows in language models.
Read more
AI Summary
Highlights
AI Chapters
Episode notes
auto_awesome
Podcast summary created with Snipd AI
Quick takeaways
OpenAI is working to release a multi-modal version of GPT-4 to showcase its cutting-edge capabilities.
Character AI is rapidly growing its user base, particularly among younger demographics in the US.
The race to develop effective AI detection tools is ongoing, with various companies and startups striving to stay ahead of AI-generated content.
Nvidia's dominance in chip design makes it difficult for chip startups to secure funding and compete in the market.
AI chip startup Inventus secures $125 million in funding with backing from Nvidia to develop network chips that improve GPU communication and utilization.
DataBricks completes a funding round of over $500 million, signaling its readiness to go public with a valuation of $43 billion.
Deep dives
SoftBank Planning OpenAI Investment
SoftBank is considering making an investment in OpenAI, building on its portfolio of technology companies.
Character AI Gaining Traction in the US
Character AI, a language model app, is rapidly growing its user base, particularly among younger demographics in the US.
OpenAI Rushes to Launch Multi-Modal LLM
OpenAI is working to release a multi-modal version of GPT-4 ahead of Google's Gemini to showcase its cutting-edge capabilities.
AI Detection Arms Race Intensifies
The race to develop effective AI detection tools is ongoing, with various companies and startups striving to stay ahead of AI-generated content.
Nvidia Dominance Poses Challenges for Chip Startups
Nvidia's dominance in chip design makes it difficult for chip startups to secure funding and compete in the market.
AI Chip Startup Inventus Raises $125 Million
AI chip startup Inventus secures $125 million in funding with backing from Nvidia to develop network chips that improve GPU communication and utilization.
DataBricks Raises $500 Million at $43 Billion Valuation
DataBricks completes a funding round of over $500 million, signaling its readiness to go public with a valuation of $43 billion.
Inside the Senate's private AI meeting with tech billionaire elites
The private AI meeting held at the Senate with tech billionaire elites emphasized the need for AI regulation. Elon Musk expressed concerns about the risks posed by AI, while other attendees also agreed that regulation was necessary. The discussion also touched on the potential risks of AI, such as self-replication and power-seeking. The meeting explored the possibility of updating regulations like Section 230 of the 1996 Communications Decency Act to hold AI companies responsible for their AI systems' outputs.
Anthropic's responsible scaling policy
Anthropic, a leading AI safety lab, has introduced an AI safety levels framework. This framework aims to address catastrophic risks posed by AI as it scales in power. The levels range from AI safety level 1 (no significant catastrophic risk) to AI safety level 4 (substantial risk of catastrophic misuse or low-level autonomous capabilities). This framework encourages companies to compete in achieving better safety standards and evaluation methods for AI systems.
EU to let responsible AI startups train models on its superclusters
The European Union (EU) has announced that responsible AI startups will be granted access to its superclusters for model training purposes. The EU's superclusters are powerful AI-capable infrastructure resources. This move is intended to encourage adherence to safety standards and incentivize AI companies to improve their safety measures. The EU is using access to superclusters as a carrot to facilitate responsible AI practices.
Biden-Harris administration secures voluntary commitments for AI risk management
The Biden-Harris administration has obtained voluntary commitments from eight additional AI companies to manage risks associated with AI. These commitments include safeguards against biochemical and radiological risks, cyber capabilities, and potential self-replication. The administration aims to ensure responsible AI practices as AI continues to advance.
Amazon requires declaration for AI-generated content on its platform
Amazon has implemented a new rule requiring self-publishers to declare whether their content sold on the platform is AI-generated. While content using AI tools for editing is acceptable, content generated entirely by AI must be flagged. This rule helps combat potential abuses and ensures transparency with AI-generated content.
Synthetic media and art: Self-publishers on Amazon must declare AI-generated content
Amazon has introduced rules on its Kindle Direct Publishing platform requiring self-publishers to specify if their content sold on the site is AI-generated. This rule seeks to address potential concerns related to content creation abuses while allowing AI-assisted editing without additional declaration. The distinction aims to increase transparency and ensure consumer awareness.
