Greetings,

At least two solid business use cases for genAI have emerged so far. One is to boost low-performing employees, the other to feed your internal data to an LLM to speed up research. How many more cases will we see by the end of the year?

• “Generative AI methods can create designs, such as small-molecule drugs and proteins, by analyzing diverse data modalities” – but the biggest challenge for genAI in science continues to be poor data quality and stewardship. - Nature
  
  
• The largest “open-source” text dataset for training models is here – just read the license carefully before you use it, because there are many limitations and requirements, making people question if it can even be called open-source. - GitHub
  
  
• If you’re wondering where to apply AI in your business, consider that it is most effective at helping low-performers – prof Ethan Mollick provides three studies which found that AI helps bad writers, boosts creativity, and elevates low-performing lawyers. - X (Twitter)
  
  
• Marketers will soon need to shape AI behaviors – as LLMs become the universal interface and personal/business AIs take over from search engines, marketers will need new tactics to boost online visibility. - VentureBeat
  
  
• If you’re building an LLM-based app, perhaps you don’t need to fine-tune the model – engineer Jessica Yao says you might achieve comparable results with few-shot prompting and retrieval-augmented generation (RAG). - Tidepool
  
  
• You can’t copyright AI art, says a federal judge in the first ruling in the US to establish a boundary on the legal protections for AI-generated artwork – potentially a sign that AI art won’t be copyrightable in the eyes of US law. - Bloomberg Law
  
  
• This server utility compresses text with LLMs – it’s an interesting use of LLMs, but I also love that its page says it “can compress (and hopefully decompress) text files”, more projects should be this honest. - Bellard
  
  
• McKinsey’s Lilli is an amazing boost for the consulting giant – with 40 carefully curated knowledge sources and over 100,000 documents, Lilli shortens research times from weeks to hours. - McKinsey
  

----

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.