Google Made A Big Move

Hello!

Hold on before we move towards today’s segment of the newsletter. I want you to grab your favourite beverage and sit in a comfy place.

As we have witnessed many recent developments in AI.

But that’s not the end of it.

Big companies are now releasing streamlined versions of their most popular models.

Following GPT-4o Mini, Google has introduced its new model, Gemini 1.5 Flash.

I’ve also got the scoop on some free AI courses and a nifty trick to make your video calls look like you're always paying attention (even when you're not).

Trust me, you won't want to miss this!

What’s Inside Today’s Newsletter

🪰 Buzz Around AI

😀 AI Recap

🪄 AI Courses

🔹 AI Learning

♨️ What’s Hot In AI

📰 AI News

Read time: 5 mins

🪰 Buzz Around AI

Apple Delays Launch of AI Features

Apple's been working on some cool new AI features they're calling "Apple Intelligence."

I was super excited about the features in the new iPhone and iPad software coming out in September.

But guess what?

According to a report from Bloomberg, we'll have to wait a bit longer.

They're aiming to roll them out sometime in October. It's like they're fashionably late to their party!

So, here's what's going on:

• Apple's new AI stuff, called Apple Intelligence, is coming out in October. That's a bit later than when the new iPhone software usually comes out.

• If you make apps, you might get to try out these new AI features soon.

• Some of the cool new Siri features? We might have to wait until 2025 to see all of them. They're not all going to be ready in October.

• Apple just made a promise to the White House. They said they'll try to make AI that's safe and trustworthy. It's like they're pinky-swearing with the government!

Why is Apple taking so long? 

Well, it's playing it safe.

Some new firms showed off AI fast. They got billions from investors. But Apple chose a careful path.

Apple wants to fix all bugs before users try the new AI. This could be smart. We've seen AI mess-ups from other big tech firms this year. Google's AI had issues. Other companies got in trouble for how they got data.

Apple knows first impressions last. How users see their AI now will stick for years. So they're making sure it's right from the start.

Google DeepMind's AI Scores Silver at International Math Olympiad

I wished I could be as smart as DeepMind's AI!

It just won a silver medal at the Math Olympics, and let me tell you, that's no small feat.

While everyone was watching the Paris Olympics, this AI was quietly making history at the International Mathematical Olympiad (IMO).

Here's how this super-hard math contest works:

* Imagine the toughest math problem you've ever seen. Now make it 10 times harder. The IMO gives you six of these brain-melters!

* It crushed 4 out of 6 problems. That's like getting an A+ on the world's hardest math test.

* It scored 28 out of 42 points. Just one more point and it would've won gold!

* This AI solved the trickiest problem of all. Only five human contestants could figure that one out.

Soon, Google will use this math genius AI to improve their chatbots. The next time you're stuck on a tough math problem, an AI tutor might be able to help!

😀AI Recap

1. Hugging Face Offers Developers Inference-as-a-Service Powered by NVIDIA NIM.

2. Llama 3.1-based AI Studio, powered by Meta, allows users to create, share, and discover custom AI characters.

3. Kling AI Video Generator is Now Available Globally to Take on OpenAI's Sora.

4. Google AI predicts long-term climate trends and weather — in minutes.

🪄 AI Courses

 NVIDIA just dropped amazing free AI courses.

They've put together six awesome courses that'll help you level up your AI game.

Let me give you a quick rundown:

🗂 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝗘𝘅𝗽𝗹𝗮𝗶𝗻𝗲𝗱

It's perfect if you want to get the basics down and understand what all the fuss is about.

🗂 Building A Brain in 10 Minutes

You can explore neural networks' learning process.

🗂 𝗜𝗻𝘁𝗿𝗼 𝘁𝗼 𝗔𝗜 𝗶𝗻 𝘁𝗵𝗲 𝗗𝗮𝘁𝗮 𝗖𝗲𝗻𝘁𝗲𝗿

A course on AI in data centres. You'll dive into GPU architectures and all that jazzy software stack stuff.

🗂 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗥𝗔𝗚 𝗔𝗴𝗲𝗻𝘁𝘀 𝘄𝗶𝘁𝗵 𝗟𝗟𝗠𝘀

The evolution and adoption of large language models (LLMs) have been nothing short of revolutionary, with retrieval-based systems at the forefront of this technological leap.

🔹AI Learning

Here's a simple guide to using Captions' new "Eye Contact" feature:

Captions now have a cool new trick! It can make it look like you're always looking at the camera in your videos, even if you aren't.

Here's how to use it:

1. Go to Captions' website. Log in or make a new account if you don't have one.

2. Once you're in, look for a button that says "Upload a video". Click it and pick the video you want to fix.

3. After your video loads, you'll see some options. Look for a tab that says "Effects".

4. In the Effects tab, you'll see something called "Eye Contact". Turn this on.

5. Wait a bit while the computer does its magic. When it's done, you can watch your video to see how it looks. If you like it, save your new video!

♨️ What’s Hot In AI

Google Adds Gemini 1.5 Flash LLM to Gemini AI

Alphabet (Google's parent company) just made a big move in AI. They're now using a faster, more efficient AI model called Gemini 1.5 Flash for their free chatbot.

Why is this important?

Running big AI models costs a lot of money and energy. By using a "leaner" model, Alphabet can keep costs down and be more eco-friendly.

What's new for users?

• The AI is better at reasoning and understanding images

• It can handle longer conversations

• It's available in 230 countries and 40 languages

• It now gives sources and links for its information

• Soon, it'll be able to look at files you upload and work with your email (if you let it)

This might not seem as exciting as brand-new AI features. But it's actually really important.

By making AI more efficient, companies can keep offering free services without going broke.

It also means AI won't use up as much energy, which is good for the planet.

Other big tech companies are doing similar things. They all want to make AI that's fast and cheap to run, but still smart enough to be useful.

📰 AI News

1. Meta has just introduced Segment Anything Model 2 (SAM 2), an advanced AI model that can identify and track objects across video frames in real time, marking an important milestone in video AI.

2. In an effort to reduce reliance on external chips for its AWS AI cloud services, Amazon released a new AI chip with performance up to 50% higher than NVIDIA's.

3. Video mimicking Kamala Harris' voice shared by Elon Musk raises concerns about AI's influence in upcoming elections

4. Over the course of 12 months, a UK hospital uses artificial intelligence for breast cancer diagnosis in order to enhance biopsy analysis accuracy and efficiency.