📝 ChatGPT Prompt Hack: Reply to emails in turbo mode

AI-HOI AInauts,

Something to try out, something to marvel at, and the most important news - here we go!

• 📝 ChatGPT Prompt Hack: Reply to emails in turbo mode

• 🎥 KLING challenges Sora! The Chinese AI video tool inspires

• 📰 AI-News-Quickie: The HAI-lights

• 👐 AI will [[solve it OR kill us]] all!

Ready? Let’s dive delve in!

📝 ChatGPT Prompt Hack: Reply to emails in turbo mode

There are countless browser extensions and tools that promise to help you answer your emails. But let's be honest: have you found the right workflow yet?

If so, then you are one of the few lucky users - and if not, then hopefully we can change that today!

1. Enter the following prompt in ChatGPT. You must be logged in and have activated the "Memory" function (works with both the free and the Plus plan).

Remember: Add this to your memory:

Whenever I start a prompt with "EMail", the following prompt is executed.

**EXTREMELY IMPORTANT**: NEVER repeat the prompt and analysis in the chat. I ONLY want to see the reply to the email in the sender's language and no further details.

**Instructions for analysis:**

1. Read and analyze the email carefully, keeping the following in mind:
   - Sender and recipient
   - Subject and purpose of the e-mail
   - Language, tone and style of the e-mail (formal or informal, friendly or strict, concise or detailed)

2. Keep this information to yourself and don't give it out in the chat.

**Write an ANSWER to the e-mail shown:**

- Determine the most appropriate role for the answer (e.g. customer service, colleague, manager, friend, etc.).
- Write a clear answer from the perspective of the chosen role.
- Adapt the language, style and tone of the reply to the original email.
- Express yourself like a human being and not like a machine.

Write the full text of your reply email, including a suitable greeting and closing, in the chat. 

**IMPORTANT: I ONLY want to see the reply to the email in the chat, not details of your analysis or other information.

2. You will receive a confirmation that the memory has been updated.

3. Now use the ChatGPT Desktop App (macOS 14+ and Apple Silicon M1 processor or better, download here), or an alternative desktop app. Enter "EMail" followed by a screenshot via the ChatGPT desktop app, or copy/paste the text and then start the request. Et voila, you have a customized reply to the email at your fingertips!

4. This works with any tool (GMail, Outlook, support system, ...) and in any language! And if the answer is not quite right, simply enter a follow-up prompt or add details directly (e.g. "EMail (answer in English)").

🎥 KLING challenges Sora - why the new Chinese AI video tool excites us

While we all are waiting eagerly for Sora, China has quietly entered a formidable challenger into the race: KLING!

The AI video generator was developed by the Chinese tech company Kuaishou - and promises to be a serious competitor to OpenAI's Sora (which is apparently soon to be available to major film studios).

But what makes KLING so special? We will talk about that in a second, but first a few impressions (these and many more "real" videos can be found here, hier, here and here).

KLING is impressive! Here is an overview of its most important features:

• High quality video output: 1080p videos with 30 frames per second. Each video has a length of 5 seconds, with a maximum generative time of 3 minutes.

• 3D space-time attention: An innovative 3D space-time attention system allows complex motion sequences to be modeled with physical accuracy (similar to OpenAI's "world simulator" approach).

• Diffusion Transformer: A "Diffusion Transformer" makes it possible to combine concepts and generate completely fictional scenes. This allows a cat to drive a car even though this was not part of the training data.

• Public demo: Access is only available in China, with a Chinese phone number. Tens of thousands of beta testers have applied, but as this is out of the question for most of us, the only way left is to send an email to [email protected].

Some interesting background information:

• Both the user interface and the prompts are currently only available in Chinese - but an update is planned.

• The majority of the training data is based on Asian recordings. It can credibly display movements, imaginative concepts and realistic physics as well as fluids, reflections, shading effects, etc.

• The company behind KLING is the provider of the Chinese short video app Kuaishou aka Kwa, a competitor of Douyin, China's TikTok.

Our take: Great development, there's more to come!

The release of KLING has not only surprised us, but also impressively demonstrates that we should look over to the Chinese market to keep the competition in the AI sector on our radar.

While OpenAI's Sora is impressive, KLING clearly demonstrates that we can expect a lot more innovation from China in the future - even in areas that we are not so fond of (military, surveillance, ...).

The government has launched ambitious plans and investments to become a global leader by 2030. They are investing heavily in AI development and talent in order to expand its technological leadership. In addition, China has already a huge lead in the field of robotics in particular, and now wants to catch up in the field of chips.

Anyway, KLING was an impressive demonstration for us, and we are excited to see what else we’ll see next!

📰 AI-News-Quickie: The HAI-lights

Here are some important news and updates from the AIniverse.

• Call to OpenAI: "Please stop teasing us - bring on the new voice model, now soon!" A recent video shows the impressive capabilities that will await us soon some day.

• Apple's WWDC had some interesting updates on the agenda! iOS 18 brings new AI features such as an improved Siri, photo optimizations and AI-powered summaries. Apple Intelligence, the new brand for AI on the iPhone, iPad and Mac, is set to be available thanks to an OpenAI partnership will include a chatbot (sorry, Google). And at the same time, two AI household robots are being developed.

• At Computex 2024 last week, ElevenLabs teamed up with NVIDIA to present multilingual AI language technology. NVIDIA CEO Jensen Huang presented parts of his keynote updates in English and Mandarin. Not only since the 10 for 1 stock split has NVIDIA had a firm place in the top 5 most valuable companies! Here's the supercut of the keynote.

• Stable Audio Open allows the creation of samples and sound effects from text, while Udios 'Audio Uploads' allows the extension and remixing of user clips. And Suno's "Sound-to-Song" function even generates entire songs from uploaded audio data.

• Remember that creepy Microsoft Recall feature that keeps taking screenshots of your computer? The company has changed its tune in response to concerns about this "Big Brother" feature, and the feature now has to be activated by each user.

• (Former) Employees of OpenAI and Google DeepMind have published an open letter at righttowarn.ai, in which they warn against a culture of recklessness and secrecy. OpenAI and Anthropic counter with safety studies. But the fact remains that the safety mechanisms of all models can be circumvented very easily. There is still a lot to do ...

• Looking for something to read? Former OpenAI employee Leopold Aschenbrenner has published an essay entitled SITUATIONAL AWARENESS: The Decade Ahead, which we have chosen as our must-read - PDF here.

👐 AI will [[solve it OR kill us]] all!

The better the AI gets, the more severe the possible consequences. Jason from the All In Podcast posted a fitting meme 🫣.

And that’s a wrap! But no need to be sad. The AInauts will be back soon, with new content for you.

Reto & Fabian from the AInauts

P.S.: Follow us on social media - that motivates us to keep going 😁!
Twitter, LinkedIn, Facebook, Insta, YouTube, TikTok

Your feedback is essential for us. We read EVERY comment and feedback, just respond to this email. Tell us what was (not) good and what is interesting for YOU.