In this issue of AI News Weekly by Keep Coding:
Elon Musk Announces Open Sourcing of AI Chatbot Grok Amid Legal Tussle with OpenAI
Elon Musk announced that xAI will make its Grok chatbot open source this week, offering it to Premium+ subscribers on X without detailing the reasons for this decision.
This move continues Musk's history of open-sourcing company technology, aiming to foster wider adoption and improvement through community feedback, similar to Tesla's open patent strategy and X's publication of its algorithmic feed code.
The decision to open source Grok comes amidst Musk's lawsuit against OpenAI, accusing it of abandoning its non-profit origins for profit, contrasting with his actions to promote open-source practices.
Ex-Google engineer charged with stealing AI secrets for Chinesefirm
Linwei Ding, a Google engineer, has been indicted for allegedly stealing over 500 files related to Google's AI technology, including designs for chips and data centre technologies, to benefit companies in China.
The stolen data includes designs for Google's TPU chips and GPUs, crucial for AI workloads, amid U.S. efforts to restrict China's access to AI-specific chips.
Ding allegedly transferred stolen files to a personal cloud account using a method designed to evade Google's detection systems, was offered a CTO position by a Chinese AI company, and founded a machine learning startup in China while still employed at Google.
Anthropic's Claude 3 chatbot claims to outperform ChatGPT, Gemini
Anthropic's latest AI chatbot, Claude 3, is making waves in the AI community with its claim to outperform OpenAI's ChatGPT and Google's Gemini. This new version promises quicker results, greater accuracy, and improved ease of use, including the ability to analyze uploaded images, documents, and other files.
Highlights:
Multiple Models: Claude 3 offers three different models - Opus, Sonnet, and the upcoming Haiku. Opus and Sonnet are currently available through the Claude 3 website and API for developers.
Enhanced Performance: In tests, the Opus model of Claude 3 outperformed GPT-3.5, GPT-4, and Google's Gemini in areas like common knowledge, expert reasoning, and mathematical problem-solving.
Advanced Capabilities: Claude 3 is noted for its "near-human levels of comprehension and fluency on complex tasks," with quicker response times and greater intelligence than its predecessors.
New Features: A significant upgrade in Claude 3 is its ability to process a variety of file types, including images, PDFs, text files, and more, for analysis and summarization.
Sam Altman re-joins OpenAI’s board
An independent investigation found Sam Altman's conduct as CEO of OpenAI did not justify his removal, leading to his re-joining the board after a failed boardroom coup last fall.
The investigation, conducted by WilmerHale, involved interviews with board members and employees, reviewing over 30,000 documents, and concluded Altman and co-founder Greg Brockman are suitable leaders for OpenAI.
Despite OpenAI's vague public summary of the investigation, it highlighted that the firing was not due to product safety, financial concerns, or development pace, but rather a breakdown in trust with the previous board, with plans to strengthen conflict of interest policies and introduce a whistle-blower hotline.
Image of the Week
Prompt: Two women sitting in an old wooden car, in the style of celebrity photography, feminine body, contemporary tapestries, Japanese prints, light amber and magenta, contemporary chicano, light brown and light amber --v 6.0
Want to get the news earlier? Subscribe to our weekly newsletter.