WednesdAI // Week 38

Even as technology transforms our world, it’s the human touch and spirit that give it true meaning.

Let’s get into it!
 
Welcome to WednesdAI – Pixel Dreams’ weekly update with top stories from the rapidly evolving world of Artificial Intelligence.

This Week’s News

Put Yourself Anywhere, Be Anything with Video to Video

Runway’s new Video-to-Video AI tool, part of its Gen-3 Alpha release, lets users take real-world video clips and apply AI-driven transformations.

This is a step up from their earlier text-to-video and image-to-video features, now allowing you to keep the original motion of your footage but completely change the look or setting. For instance, a clip of a kid running around the yard can be re-imagined underwater or on an alien planet—all with just a text prompt. For businesses, this is a shortcut to highly customized video content—less time, less effort, and no need for a Hollywood budget.

For more details, check the Runway’s official Release and coverage on Tom’s Guide.

Adobe announced their generative video platform but we’ll cover it in-depth when it releases later this year.

ChatGPT o1

OpenAI has released a preview of its new o1 reasoning model, designed to spend more time thinking through problems. Unlike GPT-4, this model excels at solving complex tasks like math, coding, and science—scoring 83% on a math Olympiad qualifier, compared to GPT-4’s 13%. It’s a stripped-down version, lacking some GPT features like web browsing or image uploads, but it shines in reasoning and safety, outperforming in jailbreak resistance tests.

 
Watch YouTuber Kyle Kabasares get the ChatGPT o1 model to write PhD code in under an hour that it took him over a year to complete:

Official release from OpenAI.
 
 

Salesforce’s Hard Pivot to AI Agents

As part of a strategic pivot towards AI, Salesforce CEO Marc Benioff introduced Agentforce, a platform for AI agents designed to automate tasks and improve customer interactions. The move comes as Salesforce faces pressure to stay competitive in the AI space, with clients like Heathrow and OpenTable already testing the tool. Benioff promises more insights at the upcoming Dreamforce conference, where AI will dominate the conversation.

Read about it at Fortune.com.

The First Violin Performance in Space

 
This moment is beautiful and profound for the human touch it represents, but the technology we talk about is what made it possible.

Enjoy this image of a SpaceX astronaut making the farthest spacewalk from Earth in 50 years.
 

Read about the performance at The Strad.
 

Videos of the Week

RunwayML’s Gen:48 contest just ended and the entries are a wonder to behold!

The images accompanying the news items in this article were generated in Midjourney using the following prompts:

Anime cartoon style. A person controlling their video reality, changing the entire visual style of their environment by simply speaking a text prompt. A cityscape where buildings, vehicles, and people shift in design–from hyper-detailed anime to minimalist line art and watercolor aesthetics–based on the person’s commands. The figure is in the center, holding a device that alters the surroundings with each new prompt, creating a dynamic visual explosion of styles around them, all set in fast-paced motion. The constant style changes showcase the fluid adaptability of the AI, with exaggerated anime motion lines adding energy to the evolving scene

Playful 3D Animation with Bright Colors. ChatGPT mini versions appearing as floating AI assistants, rapidly solving academic coding tasks for a student in a whimsical, animated world. A colorful, cartoon-like academic world where floating AI orbs zip around, assisting a student with solving PhD-level code, each task represented as glowing puzzle pieces falling into place. The student sits at a playful desk, while multiple AI mini versions hover around, each handling a different code problem, represented by vibrant, glowing symbols that move dynamically. Bright, cheerful lighting with pops of neon light that highlight the quick pace and energy of AI solving the tasks, making the scene feel lively and animated. The AI’s adaptability and efficiency are visualized in a whimsical way, with the synthetic data becoming puzzle pieces that the AI assembles into completed code in real-time

Retro Sci-Fi with a 1950s Vision. Salesforce robot agents, designed like classic retro robots, helping businesses with customer interactions. A retro & nostalgic landscape, where robots with sleek metallic bodies and Salesforce logos assist human workers in a mid-century style office. Marc Benioff stands at a classic retro control panel, overseeing the AI agents as they complete tasks autonomously, with customers smiling and interacting through vintage video screens. Soft, warm lighting like vintage film movies casting long shadows and creating a blend of vintage charm. The retro robots are given a friendly, approachable appearance, with the Salesforce interface blending seamlessly, symbolizing an idealized, efficient future

Renaissance Revival with a Space Age Twist. An astronaut in a classical pose, playing a violin in the heavens, evoking the timelessness of art while suspended in the modern marvel of space. A dramatic space environment with a focus on Earth visible in the distance, as if viewed from the Moon, where the astronaut is positioned in a moment of quiet grace, their violin’s music creating soft ripples across the vacuum of space. The astronaut is off-center, floating gracefully in the foreground, while the violin’s sound causes rippling waves that subtly blend with the curvature of Earth, creating an artistic intersection of music, humanity, and technology. Soft, golden light similar to Renaissance paintings, casting a warm glow on the astronaut’s figure, contrasting with the cold blues and blacks of the space backdrop. A sense of harmony between human artistry and the vast technological achievement of space travel, with subtle details of classical art style merged with futuristic space elements