Innovation Profs - 10/1/2024

Your weekly guide to generative AI tools and news

Sign up for half-day workshop

Oct. 16 Using Gen AI at Your Work: This completely new virtual workshop will take a hands-on approach to helping you get implement generative AI at your work. You’ll learn our process to determine quick wins, imagine new ideas and prepare for the future of AI at your company.

Generative AI News

ChatGPT's 4o-mini model just got a big upgrade – here are 4 of the best new features

Last week, OpenAI updated GPT-4o-mini, their base model for the free tier, with a number of new features: image generation, web browsing, the ability to upload images and files to be discussed, and memory, which allows certain aspects of previous conversations such as user preferences to be referenced in future conversations. Previously, these features were only available at the paid tier of ChatGPT, so this is a welcome development for ChatGPT users who aren’t willing to shell out $20 a month for more advanced features.

People are "blatantly stealing my work," AI artist complains

Jason Allen, whose MidJourney-created image “Théâtre D’opéra Spatial” won a digital art competition at the Colorado State Fair (although the judges were unaware that the image was AI-generated), has appealed the US Copyright Office’s rejection of his application to copyright the image due to its lack of human authorship. In his appeal, Allen claimed that generating the image took 624 iterations with MidJourney, amounting to 110 hours of human work. He has further claimed that the rejection of his copyright has “has put me in a terrible position, with no recourse against others who are blatantly and repeatedly stealing my work without compensation or credit.” 😂 😂 😂 

AI Startup Runway Says It’s Giving Filmmakers up to $1 Million If They Use Artificial Intelligence to Make Their Movies

Last week we reported that Runway, the company behind the text-to-video generation tool Gen-3 Alpha, agreed to a licensing deal with entertainment company Lionsgate to use their movies to train a new video generation model. This week we’ve learned that Runway is now trying to entice filmmakers to use their tools to create movies, with grants ranging from $5,000 to $1 million, as well as up to $2 million in Runway credit to work with. Runway co-founder and CEO Cris Valenzuela was clear about the motivation behind Runway’s fund: To encourage the use of generative AI tools, not necessarily their own, to create films. “We believe that the best stories are yet to be told, and that traditional funding mechanisms often overlook these new visions.”

MIT spinoff Liquid debuts non-transformer AI models and they’re already state-of-the-art

As many of you already know, the ‘T’ in ChatGPT stands for ‘transformer,’ which is a specific kind of machine learning architecture developed by Google researchers back in 2017. In addition, the LLMs that are vying for the top spot on the Chatbot Leaderboard are all based on transformers. It then comes as a surprise to learn of a new approach to building conversational AI models that is not based on the transformer architecture. In fact, such models can already outperform some transformer-based models, including Meta’s Llama 3.1-8B and Microsoft’s Phi-3.5 3.8B (which are by no means the most powerful models these companies have developed). The company behind these new models is Liquid, and their models are called “Liquid Foundation Models” (LFMs). Read more to learn about this new approach to language modeling.

Quick Hits

Tool of the week: Runway’s Gen-3 Alpha

We’ve talked about Runway’s text-to-video tool in a previous newletter. Gen-2 was upgraded to Gen-3 Alpha back in June. As mentioned above, we’re about to see some films created using the tool. Runway set aside $5 million to fund up to 100 original films using the tool. Learn more about the project.

AI-generated image of the week

If you’ve been to one of our workshops, you’ve seen this example of a mechanical cat. But Ideogram 2.0 takes it to a whole new level.

prompt: poster of a mechanical cat, technical Schematics viewed from front and side view on light white blueprint paper, Illustration drafting style, typography, conceptual art, dark fantasy, steampunk, cinematic

Get starting with Generative AI

New to generative AI? Here are some places to start…

What we found

Apple is said to be working on a smart home device that will run Apple apps like Calendar, Notes, and Home.