• AI Academy
  • Posts
  • 🧠 Remember What You Said? ChatGPT Does

🧠 Remember What You Said? ChatGPT Does

Plus Apple’s AI photo editor and goodbye Bard

Hello AI Enthusiast,

In this edition, we're introducing a fresh approach by sharing AI Academy's insights on the week's most significant and debated news. Following the News section, you'll discover a poll where you can share your feedback on this new format. Let us know what you think!

Before we jump into today's updates, we wish you all a joyful Valentine's Day! 💐

If self-love is your theme this year, why not treat yourself to something special? Coming up next month is the fifth edition of our Master in Prompt Engineering. This program has empowered our students to create AI prototypes, without coding, that have impressed bosses and clients alike. Across every sector, there's a growing desire to either boost internal efficiency or enhance the products and services offered to customers. You can be the next to stand out and innovate.

This upcoming edition will focus even more on our alumni, shining a spotlight on the incredible projects our students are building. Plus the early bird discount is about to end soon. If love is in the air, why not show some to yourself by securing your spot in the course?

Now on with the news.

News Bytes 🗞️

  • OpenAI is testing the ability of ChatGPT to remember and recall information discussed in previous chats, allowing for more helpful future conversations. Users have control over ChatGPT's memory and can choose to turn it off or delete specific memories.

    💡Our take: OpenAI's move to equip ChatGPT with memory capabilities signifies a pivotal shift towards becoming a product-focused company, introducing a significant UX enhancement. This innovation not only empowers users with privacy controls but also strategically creates a lock-in effect. As users spend time customizing ChatGPT to their preferences, they will find it increasingly difficult to switch platforms. This retention strategy cleverly positions OpenAI in a competitive stance against tech giants like Google and Apple, which already capitalize on extensive user data from services like Gmail and iCloud.

  • OpenAI CEO Sam Altman is seeking trillions of dollars in investments to address the supply-and-demand problem in the AI chip market, which currently limits OpenAI's growth. He is reportedly in talks with investors, including the government of the United Arab Emirates, and may need to raise between $5 trillion and $7 trillion for the project.

    💡Our take: Sam Altman's quest marks a strategic move with deep commercial and geopolitical implications. Amidst NVIDIA's dominance and its Taiwan-based production, Altman's discussions with entities like the UAE government aim to diversify the AI chip supply chain, mirroring efforts like the EU's $40 billion chip initiative. This strategy reflects a broader effort to mitigate vulnerabilities like those exposed during the COVID-induced chip shortages, indicating a shift towards securing a more autonomous and robust foundation for the AI industry.

  • Nvidia, the leader in AI chips, is creating a team to develop custom chips¹ for AI companies like Meta and OpenAI. This move comes as tech giants like Amazon and Microsoft and, as mentioned before, OpenAI are shifting towards in-house chip development, making personalized chips more in demand. Nvidia's entry into the custom chip field could further boost the company's growth as the demand for specialized and energy-efficient chips continues to rise.

  • OpenAI is developing agent software that can automate complex tasks by taking control of a user's device, including gathering public data, creating itineraries, and booking flights. This development demonstrates the potential of digital assistants in practical applications.

    💡Our take: This initiative propels OpenAI into a competitive landscape focused on real-world AI applications. While offering valuable opportunities for OpenAI to enhance utility outside its core platforms, it also introduces significant risks related to privacy, security, and the potential for overly autonomous AI systems, reminiscent of speculative scenarios like Skynet. The balance between expanding AI's practicality and managing these risks is critical.

  • Apple has developed a new system called MGIE in collaboration with researchers from the University of California, Santa Barbara. MGIE utilizes MLLMs, which can transform text prompts into detailed instructions for a photo editor, allowing users to make specific edits and improvements to their photos simply through text.

    💡Our take: By leveraging MLLMs to translate text prompts into precise photo editing commands, Apple not only simplifies user interaction but also capitalizes on its ecosystem advantage. Having the AI operate directly on devices where users' photos are stored offers a seamless, integrated experience, showcasing Apple's strategic use of AI to enhance user convenience while maintaining direct access to the data needed for such personalized interactions.

  • Microsoft has given its AI-powered chat assistant, Copilot, a facelift with a cleaner interface and added AI image creation and editing capabilities. Additionally, the introduction of the Deucalion AI model improves Copilot's responses, making it more efficient and versatile. The company has invested $12 million in a Super Bowl ad campaign to promote Copilot, igniting a debate about the ethical implications of AI. Our founder, Gianluca, has discussed this controversy in a TikTok video.

  • Google has revamped its chatbot, Bard, into Gemini with the release of a paid version called Gemini Advanced. Gemini Advanced uses Google's best AI model, Gemini Ultra, giving users the opportunity to try out a high-performing chatbot for free and potentially find use cases where Gemini outshines OpenAI's ChatGPT. Only time will tell if they can catch up to OpenAI's dominance.

  • Google has stopped over 170 million fake online reviews, a 45% increase from the previous year, using new AI technology. This crackdown should provide relief to local businesses affected by fraudulent reviews and aims to protect online reputations.

  • Voice actors now have the opportunity to earn passive income through AI voice cloning on the ElevenLabs platform. This gives voice actors more control and earning potential by allowing them to create an AI clone of their voice, which can be used and paid for by others. While there may be concerns about AI taking away creative jobs, this program offers a new revenue stream for voice actors and the potential for more financial stability.

  • Ikea has introduced an AI assistant in the OpenAI GPT store as part of their experimentation with AI as they shape their company strategy. However, users have found the Ikea GPT to be ineffective in providing design ideas and have faced issues with inaccurate information. Room for improvement.

Do You Like Our New Format?

We're now sharing our thoughts on the week's big news. What's your take?

Login or Subscribe to participate in polls.

This week’s glossary 📖

  1. Custom Chips: Tailor-made silicon chips designed to meet specific needs of an application or organization, enhancing performance and efficiency for particular tasks.

Educational Pill 💊

Multi-Layered Language Models and how Apple is harnessing their power

Imagine a smart AI that has devoured almost every piece of written content on the internet. This AI, through its training, learns not just to understand but also to generate language.

Apple's latest innovation, the MGIE system, leverages MLLMs to transform simple text prompts into detailed photo editing commands, simplifying the way we interact with technology. This means you could tell your device to "brighten this photo" or "crop out the background," and the AI, understanding your request and executes it seamlessly. MLLMs has the potential to make technology more accessible and interactive.

From our community 🤝

We're happy to share your thoughts from the polls in our previous newsletter. It's always exciting to see what you all think about the big subjects of AI!

On the topic of fighting misinformation as we head into election season, it turns out a lot of you—62%, to be exact—are not quite convinced that the current measures are up to scratch. You've shared that despite the efforts, there's still a way to go in tackling fake news and the like.

Then we dove into the world of AI chat assistants, where Hugging Face is shaking things up with its open-source approach. Here, a whopping 84% of you believe that open-source models, alone or combined with proprietary ones, are the way forward. This enthusiasm for innovation and accessibility in AI is exactly what we like about our community.

LOLgorithms 😂

Ok Sam, we love you, but not that much.

That is the end of our newsletter.

Remember, if your company is looking to implement AI technologies, we also offer customized corporate training.

See you next week. 👋