Spend 1 minute every day to get the latest AI information.
The content includes but is not limited to cutting-edge AI news, AI tools, AI painting, open-source projects, and learning tutorials.
The main feature of the brief is its concise description, but important information is still introduced in detail through independent posts.
Here is the latest AI information for June 29.
Cutting-edge News
1. This is the meaning of ChatGPT's birth: using technology to protect vulnerable groups.
A popular post today describes using GPT to help a child with borderline intellectual disability to prevent being scammed.
A custom GPT chatbot is set up for the child with the prompt: "I am a person with borderline intellectual disability, easily deceived or committing crimes. Your task is to remind me to consult my parents if there are any signs of this."
Then the child is told to ask GPT first instead of making decisions on their own.
For example, if the child asks: "I want to drink hot milk, can I microwave it in a plastic bottle?"
GPT would reply: "The bottle will melt. Please pour it into a bowl before microwaving, and do not use a metal bowl, only ceramic."
Weibo post: https://weibo.com/3872641610/Ol4kibuNl
Original post: https://gall.dcinside.com/board/view/?id=dcbest&no=197120
2. Microsoft launches GPT-4o integrated text-to-speech virtual digital human service.
Microsoft's virtual digital human service now integrates GPT-4o text-to-speech technology, enabling real-time interaction between digital humans and users.
Previously, Microsoft shared a case of a talking digital human customer service assisting with shopping.
Detailed introduction: https://techcommunity.microsoft.com/t5/ai-azure-ai-services-blog/azure-ai-speech-announces-public-preview-of-text-to-speech/ba-p/3981448
Case study: https://github.com/Azure/gen-cv/tree/main/avatar/interactive
Open-source Projects
1. An open-source VSCode plugin Reborn AI.
After installing the plugin, it helps you write, refactor, and improve code, supporting most mainstream language models and local large language models, and supports multiple languages, including Chinese, English, and more than a dozen other languages.
The plugin is completely open-source, allowing you to optimize and modify it according to your needs.
GitHub: https://github.com/Christopher-Hayes/vscode-chatgpt-reborn
2. An open-source tool with only 293 lines of code to parse PDFs into Markdown format: gptpdf.
Using a powerful visual large language model (GPT-4o), it almost perfectly parses PDF content, including layout, mathematical formulas, tables, images, and charts.
GitHub: https://github.com/CosmosShadow/gptpdf
Demonstration case provided by the project:
Original PDF file: https://github.com/CosmosShadow/gptpdf/blob/main/examples/attention_is_all_you_need.pdf
Parsed MD file: https://github.com/CosmosShadow/gptpdf/blob/main/examples/attention_is_all_you_need/output.md
3. An open-source tool to help you clean and structure data: OmniParse.
This tool can convert various unstructured data into structured, actionable data, making it easy to use for retrieval-enhanced generation (RAG) and fine-tuning.
Whether it's documents, tables, images, videos, audio, or web pages, it can clean and structure them.
GitHub: https://github.com/adithya-s-k/omniparse
