A big leap forward in ChatGPT-4o

ChatGPT-4o

It’s a big deal. It’s like an ultimate AI assistant or producer. Here’s why it’s making headlines.

• It’s the first actual multi-modal model we have. Much more nuanced. GPT can now reason with text, audio, and vision—all while holding human-like conversations in real time! 

• OpenAI says GPT-4o is twice as fast. Access and convenience is drastically improved. Significant reasoning improvement.
• Design capabilities: text-to-image creation feature; you can even ask ChatGPT to come up with a cool font design, entirely from your description. It can even do 3D rendering. 
• Camera features. Way ahead of anyone else. The new GPT takes notes, understands themes, and even emotions—just from an image.
• Improved capabilities to observe and tutor almost any subject and up to 50 languages in real time. ChatGPT is now better at math than human beings. 
• Voice capabilities: Improved AI assistant capabilities through direct speech interaction. Can do all your meeting summaries and notes.
• knowledge base has increased; it now has information up to October 2023.

From OpenAI on how to get GPT-4o:

We plan to launch a new Voice Mode with these new capabilities in an alpha in the coming weeks, with early access for Plus users as we roll out more broadly.

The Rundown Team Analysis

This multimodal release is a very large step forward. Voice/audio/image recognition are strong foundational tools for developers. It’s twice as fast and half the price for developers. This tool will make it easier for The Rundown team to do writing and teaching and offer better-producing and media advisory assistance from our Alisha BOT. She will have a stronger capacity to reason using ChatGPT-4o. Our prompting can elicit more sophisticated responses. What we were wondering was: why have we not received access to it at all? Why just a few partners or subgroups? The demos look a little overscripted, too.

Your personal producer or editorial assistant

GPT-4o can help you prep for an interview. Just open up the app and ask GPT 4o to help you go through almost any topic. This is going to be a game-changer in the media space. Imagine sitting with your AI production or editorial assistant, who can strategize, advise and help create your content. You can now plan, create and produce quality content on the go far quicker than ever before; all you need is your phone.

Voice variation: In another demo, GPT-4o read out an AI-generated story in different voices. And it’s responsiveness was impressive when the demo asked it change tone. And again, all in real time.

It’s also great for sound effect synthesis. All you have to do is put in the prompt and it’ll come up with an audio file of your sound effect. The creative possibilities here are massive.

GPT-4o is a great teacher

One of the best use cases for generative AI is closing the education gap. Here’s a video from Khan Academy and OpenAI showing how easy it is to help a student with their maths.

And not just with mathematics, but with languages too. You can now just take a picture of something, and GPT-4o can tell you what it is in almost any language.

There is huge potential to level the playing field here when it comes to widening access to education. And the difference here is that, as OpenAI showed in their demo, GPT-4o doesn't just have the ability to give the answer to a tough algebra question, it is now sophisticated enough to observe and understand your actions, and then give you guidance on how to solve the problem.

The Downfall of Duolingo?

One of the coolest features of GPT-4o is that you can now have live translation. This is any tourist's or language student’s dream. OpenAI released a live demo where, after speaking to GPT-4o in Italian, the app was able to translate the speech into English. In real time! Here’s the demo, or try it out yourself. You’ll see a button in the shape of headphones to the right of your chat. 

This is like Google Translate and Duolingo wrapped into one and way more powerful. And combined with the impressive teaching capabilities, it is now completely possible to learn an entire language by speaking to ChatGPT. Like your own personal tutor.

Duolingo’s share price actually dropped quite a bit this week and users across social media couldn’t help but make the connection.

I’m still learning my French on it, for now.

Impact on smaller start ups?

GPT-4o can now do what a lot of other tools can do, but better and all in one place. Why have Duolingo, Google Translate, or even AI tools like Otter? Or will this be useful for start ups to build on the OpenAI API?

Apple is expected to sign a deal with OpenAI. What does this mean?

Apple has spent time and money poaching Google’s best AI people. So why do this deal? It could mean Apple's own AI is not ready yet, especially for the annual fall drop of the new iPhone and iOS systems, and they are looking to buy some time until they are ready to roll it out.

Employers need to catch up and give their employees the right tools.

85% of Gen Z employees who use AI at work, say the tools they use aren’t provided by their employer, according to a report from Microsoft and LinkedIn. But Gen Z isn’t alone: Around three-quarters of Millennials, Gen X and Boomers say they too bring their own AI tools to work. But workers often stay quiet about their adoption of new AI technologies. Roughly half said they fear that openly using AI in the workplace will signal that they’re replaceable.