Greetings,

My son, Edward, was born last Friday and I can’t stop thinking about what a fascinating world he’s entering. By the time he’s able to use computers, it’s possible that we could already have superintelligent AI – which is why I’m glad that OpenAI is organizing a special team to ensure that when it comes alive, it’ll want to help humans and not harm us.

• Explore patterns of user-LLM interaction (like Contextual Conversation or Real-Time Progress) in this in-depth analysis based on developing an AI co-pilot. - MartinFowler
  
  
• OpenAI is dedicating 20% of its compute resources and a special team to solve the alignment problem – make sure that superintelligent AI will work for humans, not against us. - OpenAI
  
  
• Superintelligent AI is still theoretical, but some people think current AI tools are already it, like this small business that started recklessly automating processes and throwing proprietary data at ChatGPT (pro tip: don’t do that). - Blind
  
  
• Tech analyst Ben Evans says LLMs are ‘infinite interns’ that work best in cases where errors are easy to check and fix, and that we shouldn’t be afraid of our jobs being automated away. - Ben-Evans
  
  
• Regarding whether it’s OK to scrape publicly available data to train AI models, Google updated their privacy policy to state that they can use any public information online to build their AI products. - Engadget
  
  
• We must be careful with AI metaphors, writes researcher Boaz Barak, and offers his take on why popular metaphors like “stochastic parrots” or “McKinsey consultant” don’t do AI justice. - WindowsOnTheory
  
  
• Fine-tuning is a big part of customizing an LLM towards a specific task, and now you can do it for less than $20 with the Monster API for no-code fine-tuning. - Monster API

----

Found this helpful? Forward this email to a colleague.

Was this email forwarded to you? Sign up here to stay on top of the AI news.