Hey Reader,

I hope life, work and the world of AI are treating you well! We're back with 3 big things in AI.

It was a huge week for AI, so I've done my best to actually synthesize this down to just a few items – more to come next week!

By the way, make sure you're subscribed to our YouTube channel for free AI tutorials. I just posted a new one where Brian (our CTO) shows you how to connect Google Forms, Zapier and ChatGPT step-by-step to build a no-code AI client intake system.

And on to the news – which, by the way, always gets shared first in our students-only IWAI Slack Workspace.

#1 – OpenAI's Sora video model is (sort of) public

Most of us haven't actually been able to log in yet, but sometime in the very near future, we'll be able to start using Sora, OpenAI's incredible new video-generation model, at Sora.com.

I think Sora and its relatives are going to be the biggest thing in both the AI and film industries in quite a while, and I am very eager to get in there and start generating.

In fact, it's kind of funny that after opening the doors to Sora on Monday, OpenAI CEO Sam Altman said they were pausing new account creation due to "higher than expected" demand ... because I would have expected that all 8 billion humans would demand to use it right away!

Meanwhile, OpenAI had about four hours of downtime on both Sora and ChatGPT this week and it was a big deal – a reminder at how mainstream and ubiquitous these tools are becoming.

In other AI video news, Adobe wants to compete with Sora with a new Firefly AI model inside its software suite. I think they're far from actually accomplishing that, but I have become a fan of Photoshop's built-in AI tools this year.

#2 – Don't Sleep on Google AI

Google has not yet released an impressive AI product, and they've had some famous flops, like the original Gemini demo video that was quickly revealed to be significantly doctored.

Everyone who has used Gemini can attest that it's "fine" but not really the best at much of anything. For power users and AI enthusiasts, basically Gemini is third or fourth place among language models, which means there's minimal reason to ever use it.

However, Google is going big on agentic AI (i.e. AI with the ability to do computer tasks more autonomously) in the next year and is releasing Gemini 2 to support that goal. It seems like the increasing autonomy of agents is likely to be a huge story of AI's growth in 2025.

Simultaneously, another team at Google announced successful tests of a new quantum computer processor that is something like 10 septillion times faster than the ones on the market today.

I really don't like when people talk about AI or computers being incomprehensible, but I think "10 septillion years" definitely qualifies for that label.

Needless to say, it will be a long time before these hit production in a real way, but just the fact that we are dabbling with processors this fast makes it seem like all the compute-time and energy concerns that are constraining today's AI will be in our rear-view mirror in the next few decades.

#3 – 15 Times to Use AI, and 5 Times Not To (Ethan Mollick)

Brian and I enjoyed this post from Ethan Mollick, who is also the author of the excellent AI book "Co-Intelligence," on "15 Times to Use AI, and 5 Not To."

This is the one that I use frequently working with students:

Also, I felt like some of the "reasons not to use AI" were a stretch because he wanted to show that sometimes it doesn't work. Honestly, I could probably tweak every "reason not to" into a positive with an AI brainstorming partner by my side!

•••

That's all for today!

As always, if you have any questions or topic ideas for next week, just reply.

Have a great day,

– Rob Howard
Founder of Innovating with AI