The week in AI – 31st March ’24 – a quick summary

When it comes to news and updates in the world of Artificial Intelligence, this is the most interesting period. Not only because it’s new, but also because there’s so much uncertainty and no one really knows what’s going to happen. This is the best time for innovation because we just don’t know any better.

The past week was no exception. There’s exciting news as OpenAI goes to Hollywood, Elon Musk announces new updates to Grok and X (formerly Twitter), Adobe added a new Structure Reference feature to stay relevant and a few very important updates we missed from last week. Read on below.

• OpenAI goes to Hollywood: Since the announcement of Sora, OpenAI has been working with visual artists and filmmakers to test the capabilities of Sora. The results are absolutely astounding. The films are all wonderful showcases of what Generative AI can do in the world of Video. While GenAI is not new in the world of Video and tools like Pika and Runway ML have been pushing the space forward, Sora is a an interesting leap.

You can view the rest of the films here.

• Grok 1.5 (X’s AI LLM) will be released soon. As a cherry on the cake, Elon Musk has stated that all Premium users will have access to Grok soon. And of course he has also dropped the update that Grok 2, which is currently in training, is going to be better than all current models.

This basically means that we have more LLMs for us to use. The more the merrier as far as users are concerned.

• Adobe dropped their latest update to Firefly with a ‘Structure Reference’ feature. When combined with their already existing ‘Style Reference’ feature, Firefly is now up to speed with Midjourney and Magnific AI in its offering. Of course Adobe’s big pitch is that they are much more responsible in their approach to training their models. This is a huge win for all those already using the Adobe suite of products in their workflow. In a world where most other major tech giants are lagging behind, Adobe has kept pace. Kudos to the team over at Adobe. You can read their blog post about the release here.

• OpenAI is testing revenue share with builders and creators for their GPT store. This is good news for all those in the open source community. I see folks like Replicate and other such platforms will follow suit.

We’re partnering with a small group of US builders to test usage-based GPT earnings. Our goal is to create a vibrant ecosystem where builders are rewarded for their creativity and impact and we look forward to collaborating with builders on the best approach to get there.

— OpenAI (@OpenAI) March 27, 2024

• OpenAI is exploring the future of AI and Voice. They’ve been working on a synthetic voice engine since 2022 and have made significant progress. Having learned from the events of the last couple of years, they have decided not to release it to the public. They will continue testing and developing this as they figure out how to navigate a potential world of catastrophical misuse. You can read the entire article and preview their voice samples here.

---

Updates from last week that we missed: There are 2 updates that we thought were critical to cover:

• Emad Mostaque – Founder of Stability.ai – stepped down as CEO last week. Stability is the creator of Stable Diffusion – arguably the most impactful artificial intelligence related tool alongside ChatGPT. Stable Diffusion is to Image Generative AI as ChatGPT is to text. His primary reasons were that he’s just a better builder than he is an operator and that he believes in an open and decentralized development of AI. The way that things currently were, he’d be better of doing something else advancing the growth of AI, than focus on running ops, HR or anything else a CEO is required to do. There’s a kickass podcast of him with Harry Stebbings on 20VC, which you can catch here. He also did a recent podcast talking about his exit with Peter Diamandis. The full video is below. We a 100% recommend you watch this. He talks about a decentralized approach to developing AI which is the only successful future for humanity and whole bunch of other stuff that would require an entire blog post to break down.

• Suno.AI dropped an update, Suno v3, that allows you to create up to 2 minutes of music generated totally by AI. Suno has been doing some pretty amazing work and this update makes it so much cooler for creators who are looking to create music of any type, based on any mood or context to go ahead and create music. You can read about their full update here.

That’s a wrap for this week. Stay tuned for next week’s update. Also if I get a break from work, there are a few projects that are pending an update – especially some off-beat use cases for MidJourney’s Character Reference feature that’s half complete.

Do share this with your friends and family if you find it useful and drop a comment over on our LinkedIn in case you have any feedback.