Hello and welcome to the Automated, your AI tour guide.

As many have already predicted, 2025 is shaping up to be the year of Agentic AIs.

Yesterday, we got a glimpse of Meta’s vision, and today, OpenAI is stepping up with a brand-new feature that will transform ChatGPT from just a chatbot into something more like your personal assistant.

Here’s what we have for you today:

• 💃 ChatGPT Takes a Step Toward Becoming Your Personal Assistant!

• 🤷‍♂️ OpenAI’s AI Model Thinks in Chinese, and No One Knows Why!

• 🕐 How to use Google Deep Research to save hours of time.

• 4️⃣ 4 ways AI agents will be even better in 2025.

• 🤖 ChatGPT Prompt Of The Day: Uncover the hidden trends within a financial report.

 💃 ChatGPT Takes a Step Toward Becoming Your Personal Assistant! 

ChatGPT is now one step closer to becoming your fully autonomous personal assistant!

If you’re a paying user (weather Plus, Team, or Pro), you now have access to “Tasks”, a brand new beta feature that lets you set reminders and recurring tasks.

How does it work? It’s ridiculously simple.

You can tell ChatGPT something like, “Remind me when my passport expires in six months,” or “Every Friday, plan my weekend based on the weather” and like an overachieving, super-organized calendar app, ChatGPT will send push notifications to keep you on track.

Even better, it’s smart enough to browse the web on a schedule, including checking monthly for concert tickets. Just don’t expect it to buy them for you.

To get started, simply select "4.0 with scheduled tasks" from the dropdown menu in ChatGPT, and then tell it what to do—whether it’s setting reminders, fetching info, or even suggesting new tasks based on your chats.

You can manage everything through a dedicated web tab, though mobile and voice commands aren’t available just yet.

But here's the bigger picture: Tasks isn’t just a convenience feature—it’s part of OpenAI’s larger vision to introduce AI agents that can act semi-independently to the mix.

Remember when CEO Sam Altman called 2025 the big year of AI agents and how it would likely “join the workforce”? Well, think of Tasks as the opening act to a much bigger show.

While still in its early stages, this feature offers a glimpse into the future, where ChatGPT evolves from a simple chatbot into a true digital assistant, capable of handling more complex tasks.

But before you go ditching your other reminder apps, keep in mind: Tasks is still in beta, so don’t be surprised if a few quirks (or even a missed deadline) pop up along the way.

Oh, and by the way, OpenAI isn’t stopping here—they’re reportedly working on even bigger projects, like “Operator”, a fully autonomous AI agent, and something intriguingly codenamed “Caterpillar”.

Could these projects someday handle everything from analyzing data to navigating websites? It’s possible.

But for now, Tasks is a handy experiment, not a flawless replacement for your brain, but definitely smarter than the typical reminder apps, kind of like Siri or Alexa with a little extra edge.

So, if you’re a paying ChatGPT user, go ahead and give Tasks a try. Just maybe hold off on using it for wedding reminders, critical meetings, or, you know, your taxes.

Instead, think of it as a sneak peek into the future of AI assistants, with a few training wheels still firmly in place.

Want to explore this new feature in more detail? Click here to see everything Tasks has to offer.

 🎁 Referral Radar: Keep the Good Times Rolling!

Last time, our Atomic Prompts referral campaign was a hit—thanks to YOU!

The energy and engagement from our 16,000-strong community made it clear: sharing knowledge and unlocking rewards is a win-win.

So, we’re keeping the good times going with a brand-new campaign:

A book like Atomic Prompts helps you clarify ideas, share insights, and build lasting impact. It can open doors, spark conversations, and position you as a thought leader.

Plus, this referral campaign guides you step-by-step on The Automated’s journey with book creation with prompts.

Here’s how you get in:

✨Refer 1 Friend – $12 Value

Get the first 3 chapters of I, Leader—FREE! (I, Leader, retails on Amazon for $11.99—grab this leadership playbook and gain insights that are redefining how we lead and inspire.)

✨ Refer 3 Friends – $49 Value

Unlock exclusive Loom tutorial videos where I walk you through, step-by-step, how I used prompts to write and publish a book.

✨Refer 5 Friends – $100+ Value

Earn a 30-minute 1-on-1 class with me! We'll work together to craft your personalized strategy for using prompts to create your book.

📣 Let’s keep this momentum alive! Act now to claim these rewards and continue shaping the future with us.

🤷‍♂️ OpenAI’s AI Model Thinks in Chinese, and No One Knows Why!

Imagine asking an AI how many R’s are in “strawberry,” and halfway through its calculations, it starts thinking… in Chinese. Strange? Absolutely.

Welcome to the curious case of OpenAI’s new AI reasoning model, o1, which has been puzzling users by unexpectedly switching languages mid-thought.

When OpenAI unveiled its first reasoning model, o1, users were intrigued—until they noticed something odd.

Even when asked questions in English, o1 occasionally performed parts of its reasoning in other languages like Chinese or Persian.

One Redditor observed, “[o1] randomly started thinking in Chinese halfway through,” while another user on X exclaimed, “Why is it thinking in Chinese? No part of our conversation was in that language!”

So, what’s going on? Even the experts are scratching their heads.

Some suggest it could be linked to how o1 was trained. Models like this rely on massive datasets, many of which include Chinese characters and text.

Ted Xiao from Google DeepMind speculated that OpenAI and other labs use third-party data labeling services—often from China—for training high-level reasoning models.

This, he suggests, might explain a “Chinese linguistic influence” on o1’s reasoning.

Labels, essential for helping AI understand data, can sometimes skew outcomes. For example, biased labels have previously led to AI models that unfairly flagged certain dialects as toxic.

Could a similar bias be influencing o1’s multilingual musings? Not everyone is convinced, though.

Critics point out that o1 has also been spotted switching to Hindi, Thai, or other languages, suggesting the issue might be broader than just Chinese influence.

Matthew Guzdial, an AI researcher, theorized that o1 doesn’t see languages the way we do. “The model doesn’t know what language is; it’s all just text to it,” he explained.

AI processes information as tokens, breaking down text into words, syllables, or even individual letters. These tokens are tools for prediction, not cultural comprehension.

Another theory comes from Tiezhen Wang at Hugging Face. Wang speculates that o1 might be leaning on languages it associates with efficiency.

“I prefer doing math in Chinese because each digit is one syllable, making calculations crisp and efficient,” he wrote.

Could o1 be choosing languages based on what it “thinks” works best for the task?

Still, the mystery persists.

As Luca Soldaini from the Allen Institute for AI noted, “This type of observation is impossible to back up due to how opaque these models are. It highlights the need for more transparency in AI development.”

For now, the question of why o1 dreams in different tongues remains unanswered. 

Until OpenAI sheds some light, we’re left wondering why synthetic biology is in Mandarin or songs in French.

Click here to dive deeper into the mysteries of o1.

🧱Around The AI Block

• 🕐 How to use Google Deep Research to save hours of time.

• 4️⃣ 4 ways AI agents will be even better in 2025.

• ✍️ Joe Biden signs executive order to speed AI data center construction.

• 👨‍⚖️ US antitrust officials back an argument in Elon Musk’s legal battle against Sam Altman.

• 📱 Gemini 2.0 Advanced lands on mobile nearly a month after desktop.

• 🏥 AWS and General Catalyst join forces to transform health care with AI.

• 🔍 Wyze cameras will use AI to describe what they see.

🛠️ Trending Tools

• Gamma: A tool that lets you create engaging, adaptive presentations and documents with the help of AI, perfect for storytelling.

• Superhuman: An AI-powered email client designed to help you manage and respond to emails faster with advanced features like shortcuts and reminders.

• Stability AI: A platform for generating high-quality images and visuals using Stable Diffusion, great for artistic and creative projects.

• Tome: An AI storytelling partner for building narrative-driven presentations, complete with images and text suggestions.

• Timely: A time-tracking AI tool that automatically logs and categorizes your work, helping you manage productivity and optimize workflows.

🤖ChatGPT Prompt Of The Day: Uncover the hidden trends within a financial report.

If you're analyzing financial reports, spotting trends and key metrics can help you make informed decisions. You can unlock valuable insights and drive better outcomes with the right questions.

Here's a prompt that can help you with that.

I have a dataset that includes financial data for ABC Corp. 

Based on this data, could you answer the following questions: What was the total revenue last quarter? Are there any observable trends in the operating expenses over the past year?

Source

We've Compiled a List of Over 100 ChatGPT Power Prompts.

This should help streamline your interactions with ChatGPT and get the results you need more efficiently.

Best of all, It's free!

Click here to get it now.

That's all we've got for you today.

Did you like today's content? We'd love to hear from you! Please share your thoughts on our content below👇

Your feedback means a lot to us and helps improve the quality of our newsletter.