We’ve seen AI features in Adobe apps and hints at future features in Adobe MAX 2023 sneak peeks, but AI isn’t waiting. It’s finally starting to look like Snowcrash. This is a short summary of other news summaries, with more general news up top, and media-oriented sources in the bottom half of this roundup.
A song of hype and fire: The 10 biggest AI stories of 2023 by Benj Edwards at ArsTechnica runs down 2023. We went from Bing Chat the chatbot asking reporter Kevin Roose to divorce his wife and run away together, to the fever pitch of AI hype, to the market leader nearly tearing itself apart.
Another early look back is from TheAIGRID, 2024 AI : 10 Things Coming In 2024 (A.I In 2024 Major Predictions). Yet another is 10 AI Predictions for 2024 – Trends today – What I’m looking forward to next year! by David Shapiro, who overstates his hopes at times.
Related to the developments in AI are questions surrounding e/cc (effective accelerationism), a movement to accelerate technological progress as fast as possible without regulation. It’s an idea that seems irrational to a student of the idea of “technology-out-of-control” in modern social thought, yet worked in Silicon Valley for some time.
‘Everything, everywhere, all at once’ is not a wish fulfilling gem — or billionaires would have better hair and bodies. For an introduction to accelerationism, see Why Some Billionaires Are Actively Trying To Destroy The World by Joe Scott (his info box has good references). Other takes, of a sort, come from the Rick and Morty in “Billionaire Immortal Christ” and from RPI professor Langdon Winner.
The Wizard of AI (2023) by Alan Warburton is a video essay about generative AI. 99% of images and videos were created with Midjourney, Stable Diffusion, Runway and Pika to discuss and critique the legal, aesthetic and ethical problems of AI platforms. Featured at the Open Data Institute (ODI) Summit 2023, there’s background about the project at The Guardian UK.
Chaos reigned at the end of the year. This turmoil occurred amidst rumors of OpenAI employees warning of a possible Artificial General Intelligence (AGI) breakthrough, dubbed Q*. AGIs are highly autonomous systems that outperform humans at most economically valuable work, or they could run amuck by themselves. The Entire OpenAI Chaos Explained by ColdFusion summarized the events. For more see The real research behind the wild rumors about OpenAI’s Q* project by Timothy B. Lee at ArsTechnica.
There’s no time to wait, as now one wag says that Channel 1’s eerily real newscasters suggest a tumultuous future for online video. See These AI-generated news anchors are freaking me out by Kyle Orland, also at ArsTechnica.
There are a bunch more good resources on AI, but After Effects Portal is oriented to media creation. Still, occasional peeks are useful, like Sam Altman’s Q* Reveal, OpenAI Updates, Elon: “3 Years Until AGI”, and Synthetic Data Predictions by Matthew Berman.
Finally moving to the more media-centric AI news, we rely on few resources because too much is more than enough.
Here on PVC, Jeff Foster has been noting AI trends for post-production throughout 2023, as with his AI Tools Part 4: The Latest for Video, Film Production, 3D, Audio and more…
Oliver Peters recently had a nice roundup in Practical AI in Post article, this time at his Digital Film blog.
Here are a few shorter summaries of recent tools, 10 Best AI Video Editing Tools of 2023 (and how to use them) by Premiere Gal and Viral AI Video Tools for Beginners by The AI Advantage.
Onto some meatier sources, Matt Wolfe posted his 2023: AI Year in Review. His summary of the year is by month, and he has explored hundreds of AI Tools at Futuretools.io. Recently Matt also outlined an An EXPLOSION of AI Video is Happening Right Now.
Matt also recently shared his view on Pika 1.0 in This AI Tool Is The Future of Video Creation. Here’s also Pika’s company demo. Theoretically Media says that Pika Just Dropped The Mic On AI Video! and followed up with Mastering Pika 1.0 – Tutorial & Look at the New AI Video Generator! looking at video in-painting with your own source.
Some more general tools for people trying to avoid paying post-production types were discussed in 10 “Best” AI Video Generators (December 2023) by Alex McFarlan and This AI Fully Automates & Creates ENTIRE YOUTUBE VIDEOS! by MattVidPro AI on invideo AI. I wonder if it will effect fishing expeditions by producers looking for information so their children can finish the production!
Theoretically Media has been a great guide to the explosion of AI media tools. He recently posted AI Filmmaking Just Leveled Up! on the Motion Brush in Runway ML’s Gen-2 and noted that Hugging Face has a LCM (Latent Consistent Models), a hybrid of text and painting in real time with a model. Theoretically Media also posted AI Video Keeps Getting Better! Motion Brush is Here! describes the new Runway ML Gen 2 feature for “in-painting,” as well as a Style artistic presets feature. The Motion Brush defines an area to auto-animate in a direction.
Theoretically Media also shared The Future of AI Video Has Arrived! (Stable Diffusion Video Tutorial/Walkthrough). He tested out Stable Diffusion Video using different tools like Pinocchio, Hugging Face, and Replicate, and showcased the video quality, motion controls, and 3D understanding of this new model. He also covered Final Frame, a web tool for extending AI generated videos by chaining clips from tools like Stable Diffusion Video. Funny how the future “arrives” every day.
Ai Flux posted tidbits you might have a hard time tracking down otherwise. See for example, Apple Goes ALL IN on AI with M3 Silicon and Midjourney V6 (BETA) is out! 👀😮 Image 2 Video Coming Soon?
According to Artificial Ignorance: “Apple recently released a new paper that could suggest how the company is thinking about AI. In a nutshell: LLMs optimized to run with far less memory (e.g., on an iPhone).“
By the way, it seems the training of AI data sets is still relying on copyright images. See the thread originated by Reid Southern on the AI Joker image from Midjourney v6. The money guys think it would cost too much to fix and argue that their use counts as “fair use”!
Also, Stanford researchers found problematic content in another dataset, including over 1000 images that were child sexual abuse material.
Magnific AI is favored by many, so See The Best AI Upscaler Makes GTA 6 Look AMAZING! by Theoretically Media. For MattVidPro AI, The BEST AI Upscaler I’ve Seen to Date! is also Magnific AI, with Krea AI getting a nod too. I’m still using Premiere Pro or Red Giant, and haven’t compared the tools.
If you suspect that Firefly 2 might not be an optimal solution for the extra money, check out What AI Image Generator Should YOU Be Using?? by Matt Wolfe. It’s a detailed comparison.
Check out also Theoretically Media’s Is Meta’s New AI Image Generator Good? Meta trains on Instagram and Facebook uploads (they own your work). For now, Meta’s tool is free, unlimited, and US-only. Also check out AI Image Shootout: Midjourney vs Meta AI by Tony & Chelsea Northrup.
MattVidPro AI shared Almost UNBELIEVABLE! – The Best AI Voice Changer I’ve Ever Heard (Eleven Labs). He also posted about the demo for Meta Audiobox in Now This is Scary Good! True AI Text to Sound! Create/Edit ANY Sound!
Curious Refuge is a useful website with YouTube AI filmmaking news roundups, and their favorite tools (a good overview of the year itself). Recent news videos include The Most Realistic AI Video Tool Yet! on Gaussian Splatting, OpenAI DALL-E 3, and more, as well as Google’s AI Video Generator Beat Runway in Side-by-Side Test!
They also seem to have a podcast; see for example The Future of Filmmaking with Tim at Theoretically Media.
Thanks to deepak pal for the main image!
Ω