gen-ai

ChatGPT stands zero chance on WhatsApp unless they change this quickly

By / Last updated on January 3, 2025

ChatGPT is now on WhatsApp, but right now it’s too obvious it stands no chance against Meta AI.

No web access is a big one.

For starters, they’ve got a cool branded number, but that can never compete with how easy Meta AI is to access.

And you still have to create a new contact to actually add the number.

May not seem like a lot of work but we humans can be pretty lazy and a tiny bit of friction like this is all takes to stop a lot of us from ever giving WhatsApp ChatGPT a shot.

I know it took much longer to try it out compared to Meta AI that was just there. One tap of a button.

When you clear your Meta AI chat it’s so easy to start a new one with the same button — but with ChatGPT you have manually search for it in your contact list like for any other contact.

They just can’t compete with this native integration. It’s also a major advantage Apple and Google have over every other AI-obsessed company. No AI will ever be able to integrate deeply with iOS and Apple ecosystem as much as Siri + Apple Intelligence. Even they’re better.

Even within chats they also couldn’t match up in usability.

Read receipts — seems small but that feedback can go a long way. It’s a big reason why me and lot of people keep them turned on.

ChatGPT doesn’t have them like Meta AI.

Meta AI even has a loading indicator when processing your messages — probably a native feature only accessible in that chat.

And when you finally get the response from Meta they will be much more up-to-date than what ChatGPT gives you.

Meta AI can give answers straight from the web.

Right now it’s powered by Llama 3.2, and they will only keep upgrading the model and built-in cutoff point.

Meanwhile ChatGPT on WhatsApp actually uses the outdated GPT-4 model — which means it’s knowledge cut off is still in April 2023.

And it doesn’t even have access to the web to compensate this.

And they only GPT-4 so they’ll be no image generation like in regular ChatGPT.

Meta AI can even join group chats and give useful info when you tag it.

Native first-party features like this will make really hard for OpenAI to overthrow Meta AI in home territory.

But they still have a real chance in this.

They do have the calling feature which Meta doesn’t — though I doubt most people will use it.

But like if they upgrade the model on WhatsApp to GPT-4o, they’ll be able to work with images and audio — a great advantage of Meta AI that doesn’t even let you upload anything.

They’re definitely seeing the advantage of being so close to users directly in their favorite chat app where they have all their friends and family.

Especially when there are over 2.7 billion of those users.

Google really destroyed OpenAI and Sora without even trying

By / Last updated on December 23, 2024

Just when Sam Altman thought they were far ahead of the competition with Sora…

Google came along and destroyed them with their new Veo 2 video creation AI model.

Has it even been up to a week? But Sora is already outdated and no one cares.

Sora what are those flamingos doing??

I’m not so sure about this skateboard motion cause how of fast it is, but it does look pretty realistic. And the camera motion is so dynamic and fluid. And look at the spectators in the background.

This one went viral across the internet:

Veo’s video is just far far better in every way.

Realistic motion — the tomatoes actually get sliced and separated. And look how they move realistically according to the slicing motion.

Better lighting — just look at the shadows of the hand and knife and how they move with the motion.

Actually makes sense — if the man in Sora sliced his fingers off at least the video would have been consistent, lol.

Veo 2 can produce videos in stunning 4K resolution — Sora can only produce in 1080p. And it follows the prompts far more accurately than Sora.

With Veo 2 you can even specify camera angles, lens types, and cinematic effects to achieve the exact aesthetic you need.

Look how complex this scene is — so many moving objects:

Veo 2 is already giving us up to 2 minutes video yet Sora can’t even go past 20 seconds.

With 2 minutes I can easily make a short video or a powerful TikTok video with nothing but my mind. What is 20 seconds good for? Unless you’re making short video clips for a larger video.

And now Google is already possibly planning to integrate Veo 2 into YouTube Shorts — promoting and making it a real tool people use to create actual content.

But OpenAI is still struggling with keeping up with high Sora usage.

But I just saw that they’ve added a new link in ChatGPT to promote it as a companion, so they’re clearly not messing around.

Many of us had been mocking Google for Gemini’s apparent inferiority compared to GPT.

But look, now they’ve dropped a far more powerful Gemini with support for multimodal agents — Gemini 2.0.

Just look at what the new Gemini-powered Project Mariner can do in the browser:

And they’re the ones laughing now with the most advanced video creation AI in the world — at least till next month.

Let’s just keep watching the AI race to see where it leads.