ChatGPT on WhatsApp now supports Images and Voice Messages

Hello, AI Enthusiasts!

Welcome to FavTutor’s AI Recap! We’ve gathered all the latest and important AI developments for the past 24 hours in one place, just for you.

In Today’s Newsletter: 😀

• ChatGPT on WhatsApp now supports Images and Voice Messages

• DeepSeek is Censored even if you run it locally

• AI models struggle with complex historical questions

OpenAI
💭 ChatGPT on WhatsApp now supports Images and Voice Messages

OpenAI has expanded ChatGPT's integration with WhatsApp, adding several features that bring it closer to the full browser experience. You can now send images and voice notes to the WhatsApp AI chatbot, which will respond with text. The ChatGPT in WhatsApp experience is free for all users.

Insights for you:

• You can now share images with ChatGPT on WhatsApp and ask follow-up questions for further interaction.

• Users can also send audio messages instead of typing their queries. The chatbot will analyze these audio snippets and respond with text.

• To start using ChatGPT on WhatsApp, users must save 1-800-CHATGPT (1-800-242-8478) as a contact on their smartphone and relaunch the app.

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

DeepSeek
😐️ DeepSeek is Censored even if you run it locally

There are online claims that operating DeepSeek's AI model locally would bypass its censorship features. However, the Wired Investigation has disproven it. The built-in bias stated in the programming code operates continuously throughout all software operational environments.

Insights for you:

• When Wired hosted DeepSeek’s model on Together AI, it didn’t outright refuse to answer a question, but it still exhibits signs of censorship.

• New evidence demonstrates that censorship runs throughout DeepSeek's knowledge database, on both the application and training levels.

• During its operation, the AI technology received commands to ignore mentioning any sensitive historical events, while prioritizing positive aspects of the Chinese Communist Party.

AI Research
🗺️ AI models struggle with complex historical questions

A new study conducted by a team from the Complexity Science Hub (CSH) has found that artificial intelligence (AI) systems are failing to respond to complicated historical queries. They created a new benchmark, Hist-LLM, to evaluate 3 Biggest LLMs: GPT-4 Turbo, Llama, and Gemini, on their accuracy in answering historical questions.

Insights for you:

• Researchers created the Hist-LLM benchmark that assesses the accuracy of responses against the Seshat Global History Databank, a rich repository of historical data.

• The results show that even the top-performing LLM, GPT-4 Turbo, managed to get only around 46% right.

• The study also noted that OpenAI and Meta models performed poorly on questions about certain regions such as sub-Saharan Africa, suggesting possible bias in their training data.