Greetings,

One of the things that fuels my optimism about generative AI is seeing cool and useful projects built by passionate developers who make them available for free – like the two GPT-powered apps I mention below. It tells me that there’s excitement about building with GenAI, and when developers are excited, a lot of disruption can happen in a short period of time.

• If you have an idea for a web app and want to build a simple MVP, try telling this GPT-powered web app generator to build it for you. - GPT Web App Generator
  
  
• When you’re struggling to get started on a task or find ideas for what to cook, try one of these small, simple, GPT-powered productivity tools. - Goblin Tools
  
  
• If you’re building AI-powered product features, this is a must read – Product Security Engineer Rami McCarthy provides a very detailed guide about building features based on AI APIs. - tl;dr sec
  
  
• Due to problems like data contamination, robustness, and flawed benchmarks, it’s hard to assess whether AI systems are intelligent – to study this properly, there needs to be more transparency about how models are trained. - Science
  
  
• On the topic of transparency, here’s an archived Twitter thread with leaked details about how GPT-4 was built – the thread was taken down due to a copyright claim, so you should definitely NOT open this. - Archive.Today
  
  
• If you didn’t open it, there’s also an article about GPT-4’s engineering secrets, however it’s behind a $500 paywall so try to get your company card to pay for it (or just read the intro, which by itself is great). - SemiAnalysis
  
  
• Claude 2 is a much cheaper GPT-4 competitor from Anthropic, with a 100K context window, strong reasoning skills, and particularly good ability to write long-form content (sadly, it’s only available in the US and UK for now). - Anthropic
  
  
• But wait, another competitor enters the ring – Meta is getting ready to release a commercial version of their open-source LLaMa model for text, image, and code generation. - ZDNet
  
  
• Meanwhile, the underdog rises again – Google’s Bard just got a huge update, it’s finally available in Europe and several other locations, you can prompt it with images, and it can give audio responses. - Google

----

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.