Innovation Profs - 10/8/2024

Your weekly guide to generative AI tools and news

Innovation Profs
October 09, 2024

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

OpenAI rolls out Canvas, its newest ChatGPT interface

Since the release of Claude Artifacts this past summer, none of Anthropic’s LLM rivals have released a comparable tool. Until now. Late last week, OpenAI released Canvas, which looks uncannily similar to the Artifacts interface: a separate window, aptly called a “canvas,” opens alongside the normal ChatGPT interface, showing the production of standalone content as requested in a user’s prompt. Unlike Claude Artifacts, this content is directly editable with a number of shortcuts appropriate for the given task (e.g. writing shortcuts and coding shortcuts). In addition, in a coding-based canvas, ChatGPT will make inline suggestions for improving one’s code, functioning as more of a coding assistant compared to Claude Artifacts. Let’s wait and see if these features make their way into Artifacts in the near future.

Meta unveils advanced video-creation AI

Last week, Meta announced Movie Gen, a yet-to-be released text-to-video tool that will have features not available in any alternative AI video tool, including the ability to make longer video clips and more precise editing features such as replacing certain objects in videos with other objects: “In one example, a lantern is converted to a bubble, while another sample replaced a Quest-style VR headset with steampunk goggles.” We’re still in an early phase for AI video generation, with Movie Gen joining OpenAI’s Sora and Adobe’s Firefly Video Engine as tools that have been announced and demoed by select users but not made available to the public.

Move over, Midjourney — Flux just got a huge upgrade for generating AI images

Flux, the open source image generation tool made by Black Forest Labs (and which powers the image generation capabilities of Elon Musk’s LLM Grok), will soon be upgraded to Flux 1.1 Pro, a version set to rival top image generation tools Midjourney 6.1 and Ideogram 2. The tool is reportedly six times faster than the previous version of Flux, but also comes with improvements in “image quality, prompt adherence and diversity.” The tool will be available through the website Freepik, as well as via API calls.

Quick Hits

I write about AI for a living — and NotebookLM is the most exciting tech to arrive since ChatGPT

The future of search is here!

www.tomsguide.com/ai/i-write-about-ai-for-a-living-and-notebooklm-is-the-most-exciting-tech-to-arrive-since-chatgpt

Nvidia just dropped a bombshell: Its new AI model is open, massive, and ready to rival GPT-4

Nvidia has released NVLM 1.0, a powerful open-source AI model that rivals GPT-4 and Google’s systems, marking a major breakthrough in multimodal language models for vision and text tasks.

venturebeat.com/ai/nvidia-just-dropped-a-bombshell-its-new-ai-model-is-open-massive-and-ready-to-rival-gpt-4

ChatGPT is the most searched AI tool by far, but number two is surprising

A new study looks at the AI tools people are searching for. While the expected name is at the top, the rest of the list is less predictable.

www.zdnet.com/article/chatgpt-is-the-most-searched-ai-tool-by-far-but-number-two-is-surprising

Cohere just made it way easier for companies to create their own AI language models

Cohere unveils advanced AI fine-tuning tools for enterprises, revolutionizing customization of language models with improved performance and efficiency.

venturebeat.com/ai/cohere-just-made-it-way-easier-for-companies-to-create-their-own-ai-language-models

Hacking Generative AI for Fun and Profit

The Sundai Club meets once a month with a goal of pushing the limits of generative AI. Earlier this year, its members built me a handy tool for journalists.

www.wired.com/story/sundai-club-generative-ai-hackathon-group

AI-generated images can teach robots how to act

Gen AI models aren’t just good for creating pictures—they can be fine-tuned to generate useful robot training data, too.

www.technologyreview.com/2024/10/03/1104958/ai-generated-images-can-teach-robots-how-to-act

AI Shocktember: Agents, Lawyers Reveal New Fight as Ground Suddenly Shifts

Rapid changes rock talent lawyers, agents and execs as big money pours in, Newsom legislates Hollywood's future, and studios get 'exactly what they wanted'

theankler.com/p/new-ai-deal-anarchy-digital-replica-newsom-lionsgate-runway-

Tool of the week: Claude Artifacts

One of our favorite newer AI features is the ability to build Artifacts in Anthropic’s Claude. Artifacts are standalone content in a dedicated window separate from the main conversation. Artifacts make it easy to work with significant pieces of content that you may want to modify, build upon, or reference later.

For example, if you ask Claude to create a game, it will create that game as playable right next to your prompt, like this:

AI-generated image of the week

Most of our AI image generation tools no longer will make images of celebrities. Ideogram, however, still does. So we decided to revisit our Taylor Swift and Travis Kelce image from last year. The problem? Ideogram doesn’t know who Travis Kelce is.

prompt: taylor swift and travis kelce eating bbq

Get starting with Generative AI

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

What we found

Generative AI video tool KlingAI now has a lip-sync function. Users can upload audio files to their videos, and the character’s lips will move to sync with the audio. See example here.