If the product matches the demo, video editing is about to be more efficient than ever…
Adobe recently teased new AI features coming to Premiere Pro.
Chores like erasing or adding objects and extending backgrounds, which we’ve already seen for images, will soon be available for video.
For example, the Generative Extend tool can pause a character’s face for a little longer by adding more frames to your video. Pretty mind blowing stuff.
Supposedly these features will be released sometime this year, so if you’re making video content or ad creatives, keep your eyes open.
Now for the macro shot: Stanford University published an AI Index Report, summarizing its capabilities and usage.
Main takeaways include the following:
- AI beats humans at simpler tasks like image classification, but falls short in complex problems such as mathematics.
- Training costs are incredibly high. OpenAI spent around $78M just to train GPT-4. It wouldn’t surprise us if those costs eventually get passed to users…
- There are no standardized evaluations for language learning models (LLM)s, which makes rating their quality and capability difficult, to say the least. Can you really say GPT-4 is better than Claude 3?
And of course, AI makes workers more productive and allows them to produce higher-quality work—with human oversight, of course.
But regulations in the US are rolling out fast, too, so make sure you stay on top of what’s allowed and what isn’t.