AI-mazing advancements: Your high-tech, low calorie easter basket

With the current wave of AI innovation and a long weekend on our doorstep, it feels like the right moment to drop another "5 Things" list, showcasing the bits of AI wizardry that have recently caught my imagination.

Strap yourselves in.

1. We’re getting a first look at videos generated by OpenAI’s Sora

Text-to-video generator Sora captured everyone’s imagination back in February this year with its truly life-like output.

At the time OpenAI stated that it was only available to red teamers to assess critical areas for harms or risks - but at the same time granted access to several visual artists, designers and filmmakers.

Yesterday, we got a first look at some of the short films they have created.

Shy kids (one of those creators) said, “As great as Sora is at generating things that appear real - what excites us is its ability to make things that are totally surreal”.

Take a look at their film below.

2. Simulating warehouses

This is one for the commercial agents amongst you … if you haven’t heard of the ‘Omniverse’ or ‘digital twins’, you have now.

In the video below, you’ll see an incident occurring in a warehouse that is almost completely rectified through a digital twin and AI - the idea being that you can simulate pretty much any situation in a building that’s even built.

I wonder how long before residential property managers can take advantage of this type of technology.

3. Improve any of your ChatGPT prompts

If you’ve been reading my posts for a while, you’ll know the situation with any large language model like ChatGPT is it’s a situation of the old 90s computer adage of “garbage in equals garbage out”, which is why prompt engineering has become a thing.

Yesterday, I noticed a GPT called Prompt Improver on the GPT store, so I decided to give it a test run.

I chose the “Build a new prompt from scratch” option, and then asked for a video script.

We went from an extremely short, vague prompt to an extremely long and detailed one; but if you have a use case where you are not quite getting the results you’d hoped for, this is worth a try.

Here’s the link (you need to be on ChatGPT 4 to access the GPT store!)

4. It takes 11 weeks to get productive with AI

We’ve just finished the first virtual AI-Powered Agents Live, and while all of our students see the productivity benefits that AI can bring, it does take time and practice to remember that you have a personal assistant—and it also takes time to learn to use it well!

On Wednesday, Microsoft released new research on exactly how long it takes for AI to have an impact, based on the workflows of 1,300 Copilot for Microsoft 365 users.

Here’s the TL;DR:

1. If you are just venturing into AI, do something easy that will save you 11 minutes a day—things like using AI to take notes in meetings, summarise long documents, or YouTube videos.

2. Stick with using AI at least until you reach the 11-week mark since it takes at least that much time to build new habits.

Get free access to Gemini 1.5

Some of you know I’ve been hugely impressed with Google’s Gemini, especially compared to those not-so-clever previous versions called Bard.

If you want to test the next version of Gemini, you can do so for free via Google’s DeepMind.

1. Simply click on the link above and scroll down until you see a video of Sundar talking about AI in general.

2. Under that, there is some info about Gemini 1.5 along with a button that says Try Gemini 1.5

3. Then click on the button that says “Get Started” and you should be good to go.

So there you go - 5 amazing things that are happening in the world of AI this week.

But it’s Easter/a long weekend, and most of you reading this might be taking on work duties of a different kind, such as ‘Chief Egg Decorator’ or ‘Chocolate Distribution Specialist’.

So here is a tip from this time last year: if you are short on time and need to make some kids happy over the holidays, don’t be a bunny - ask the robot for help.

Stay safe, and Happy Hunting 🐰