Open Letter Sparks Debate, Rewind Revolutionizes Recall, Auto-GPT Goes Autonomous and Meet MiniGPT-4!

Author

Claire Xue
July 10, 2023

AI Generated Image

It's Claire from Moodelier 👋 your go-to source for all things generative AI and creativity!

Welcome to your weekly dosage of all things AI, tech, and the magic they bring to our everyday lives. This week we're diving headfirst into an open letter that has the AI community buzzing, but not before we explore some other fascinating developments in the tech sphere. 🚀

⬇️ Read on to find out more ⬇️

✨ Spotlight of the Week : The Open Letter 📜

Our main spotlight this week is focused on an open letter that's stirring some serious discussions in the AI community. The bone of contention? The increasing use of AI-generated illustrations in the media and publishing industries. The authors outline a number of concerns, such as:

  • The Potential Extinction of Traditional Illustration: The fear is that the unique bond between human writers and illustrators might be threatened by AI tools that can generate detailed, low-cost images mimicking human-drawn illustrations.🎨

  • Copyright and Compensation Issues: AI-art generators use vast datasets containing millions of copyrighted images, allegedly acquired without consent or knowledge of the original creators. The authors go as far as referring to this practice as "the greatest art heist in history".🚫

  • Quality and Originality of Art: The argument here is that AI-produced art lacks the insight, wit, and originality inherent in human-created art. According to the authors, AI can only create variations of existing art, leading to a gradual decline in visual culture.💭

  • Economic Impact: While the livelihoods of illustrators are at risk, the companies developing this AI tech are reaping huge profits. The authors suggest that the wages of living artists are being sacrificed for Silicon Valley's investment in AI.💰

The letter concludes with a rallying cry to artists, publishers, journalists, editors, and journalism union leaders to stand against the use of AI-generated images in lieu of human-made art, and to uphold the laws and values that protect intellectual property rights. Join the conversation and share your thoughts.💡

🔎 AI Tech Discovery - Rewind 🔄

Next up is Rewind, an AI tool aiming to become your "personal search engine". Primarily focused on enhancing personal productivity and recall, its main features include:

  • Recording and Searchability: Records everything you see, say, or do on your Apple Silicon-powered Mac and makes it all searchable.✅

  • Privacy: All the data it records is stored locally on your device, emphasizing data privacy.🔒

  • Instant Access and Reduced Need for Note-Taking: By recording all your activities, Rewind allows you to instantly access any moment in your digital history without needing to manually take notes.📖

  • Meeting Recordings: Rewind can automatically record your meetings, making them searchable later.📈

  • AI-Powered Queries: The integration with OpenAI's GPT-4 allows users to ask Rewind about anything they've seen, said, or heard.💡

  • Data Compression: To store years of recordings locally, even on devices with small hard drives, Rewind uses advanced data compression techniques.📦

While it promises to help users manage their digital lives better, it also raises significant questions about data privacy and consent.

🤓 AI Project Highlight - ChatGPT app for iOS 🚗

OpenAI has just launched a free iOS app for ChatGPT. It's cool because it syncs chat history with the web, supports voice input, and works on both iPhones and iPads. And guess what? They're planning to roll out an Android version soon. For now, it's available in the US, but other countries will get it in the coming weeks​!

Before this, the best way to access OpenAI’s language models on mobile was through Microsoft’s Bing app. With the launch of the official ChatGPT app, it's likely to attract some of these users away from Microsoft. Plus, it should help people avoid those spammy fake apps claiming to offer access to the chatbot on mobile

👀 AI Model in Focus - MiniGPT-4 🔬

Last but not least, meet MiniGPT-4, a model designed to enhance vision-language understanding. It builds on the abilities of GPT-4, demonstrating impressive capabilities like detailed image description generation, website creation from handwritten drafts, and even teaching cooking based on food photos. Best of all, it's highly computationally efficient, as it only requires training a projection layer using approximately 5 million aligned image-text pairs.

It's a wrap for this week! Don't forget to share your thoughts on the open letter debate and subscribe to our newsletter to gain access to The AI Stuff Creators and Artists Should Know .💌 

Love our newsletter or have ideas for improvements? We'd love to hear from you! Schedule a call with me and share your thoughts.

Stay in the creative mood and harness the power of AI,

Moodelier and Claire 🌈✨