Innovation Profs - 4/8/2025

In this week’s newsletter

• This week’s top Gen AI news

• In the classroom with Innovation Profs

• Tool of the week

• Image of the week

• What we found

Upcoming Innovation Profs Events

AI in Law and Policy Lunch Club
When: 12:30-1:30 p.m. Central April 9
Where: Virtual on Zoom
Cost: $20
The aim of this session is to give a broad overview of a range of legal issues pertaining to generative AI, ranging from the local level, informed by the latest developments at the intersection of AI and law/policy. Sign up now.

Copilot Deep Dive Workshop
When: Noon-3 p.m. April 11
Where: Virtual on Zoom
Cost: $150
Do you have access to Microsoft 365 Copilot, but you have not had time to explore everything you can do with the AI tool? This hands-on workshop is for you. We will walk you through how to use Copilot in Outlook, Teams, Powerpoint, Excel and more. Sign up now.

Intro to Vibe Coding Lunch Club
When: 12:30-1:30 p.m. Central April 23
Where: Virtual on Zoom
Cost: $25 early bird price through April 11
Attendees will learn about vibe coding, a new trend in generative AI where people are able to code projects with little to no coding skills by simply describing what they want to create. Sign up now.

Latest Gen AI News

Meta releases new AI model Llama 4

Llama 4 Scout and Llama 4 Maverick, Meta’s “most advanced models yet,” were released on Saturday; a third Llama 4 model, Llama 4 Behemoth, is still being trained. According to a release by Meta, these new models will be “the best in their class for multimodality,” which refers to their ability to process information across a range of domains (i.e., text, images, audio, video). Like previous Llama models, the new models are open source.

Midjourney releases V7, its first new AI image model in nearly a year

Look out, ChatGPT with Images, Midjourney just released it’s latest version, V7. According to the company, V7 “is our smartest, most beautiful, most coherent model yet.” V7 has two versions, Turbo and Relax, and can also be used in Draft Mode, which renders images ten times faster at half the cost, with the trade-off being that these images are of lower quality (but can still be upscaled). Additional features, such as image upscaling and retexturing, will be added to the tool in the coming months. In order to access V7, users need to rate around 200 images to build a Midjourney personalization profile.

OpenAI Plans O3 and O4-Mini Release Before GPT-5, Altman Says

Previously, OpenAI had announced plans to shelve reasoning models o3 and o4-mini, instead assimilating them into GPT-5. Late last week, however, Sam Altman announced that these two new models will be released on their own after all, claiming that OpenAI “found it harder than we thought it was going to be to smoothly integrate everything” into GPT-5. Altman continued, “we want to make sure we have enough capacity to support what we expect to be unprecedented demand.” According to Altman, these new models should drop “probably in a couple of weeks.”

Quick Hits

In the Classroom with the Innovation Profs

Follow along as we teach Drake University’s first ever generative AI course.

This week’s topic: Gen AI Productivity Tools
Last week: Vibe Coding (note: next week, we will share what our students made with vibe coding)

Lesson: This week we will be introducing students to tools built to make you more productive at work, including the Microsoft 365 Copilot suite of tools and Google Workspace. Students will explore creating first drafts of documents, analyzing data and creating presentations using generative AI.

Your reading: Learn about Microsoft 365 Copilot and Google Workspace.

Homework: Our students actually have no homework this week! But we would love to see you at our Friday workshop in using Copilot at Work.

Tool of the week: Tiktokenizer

​Tiktokenizer is an online tool designed to interactively visualize how OpenAI's tokenizers process text. Developed by the GitHub user dqbd, it allows users to input text and observe its tokenization, providing insights into token counts and structures.

AI-generated image of the week

Midjourney 7 is here. Midjourney has long been our go-to tool for creating generative AI images, and the latest version (introduced as an alpha version that you have to personalize to unlock), includes a new, faster draft mode. Here’s what they say about the new model: “It’s much smarter with text prompts, image prompts look fantastic, image quality is noticeably higher with beautiful textures, and bodies, hands and objects of all kinds have significantly better coherence on all details.”

Prompt: A dreamlike, digital oil painting of an enchanted forest at night, filled with fantastical elements. Oversized, bioluminescent mushrooms in shades of blue and violet glow softly on the forest floor. Trees with twisted, curling branches stretch toward a star-speckled sky. Suspended in the air is a massive, translucent jellyfish glowing with soft purples and greens, its tentacles drifting like mist. Fireflies leave glowing trails, and a small stream reflects the strange lights above. The color palette should be rich and vibrant, with heavy brushstroke textures visible in the foliage. The overall mood is mysterious and serene, evoking a sense of awe and magic.

What we found

Pika Labs lets you take better control of your generative AI videos by setting five different reference frames (images). PIka will turn those five images into a video of up to 25 seconds.

Create your own at the Pika website, or on the iOS app.