- Innovation Profs Newsletter
- Posts
- Innovation Profs - 4/1/2025
Innovation Profs - 4/1/2025
Your weekly guide to generative AI tools and news
Innovation Profs
April 01, 2025
In this week’s newsletter
Upcoming Innovation Profs Events
April 9: AI in Law and Policy Lunch Club - Generative AI is still in a “wild west” phase with respect to US legal policy, as many legal issues about copyright, fair use, and ownership remain unresolved, even as courts have issued significant rulings in a number of lawsuits. This presentation will give a broad overview of these issues, informed by the latest developments at the intersection of AI and law/policy.
April 11: Virtual Copilot Deep Dive - Are you ready to fully explore how to get the most out of Microsoft 365 Copilot at work? Sign up for our three-hour virtual workshop to get hands-on and try out Copilot in Teams, Excel, Outlook, Powerpoint and more. Sign up by 5 p.m. on April 2 to get the early bird rate of $150/person.
Latest Gen AI News
OpenAI rolls out image generation powered by GPT-4o to ChatGPT
Last week OpenAI replaced the image generation capability in ChatGPT, previously powered by DALL-E 3, with a significantly upgraded model that has vastly surpassed all other currently available image generation tools. Variously dubbed “Images in ChatGPT,” “4o Image Generation,” and “ImageGen,” the tool is available as part of the multimodal LLM GPT-4o (remember, the “o” stands for “omni,” with the thought being that GPT-4o would be able to process data in all modes rather than merely in some). Key features of this improved tool are improved binding, which refers to how attributes are connected to objects in prompts are preserved in the images, improved text-rendering, transparent backgrounds, and the ability to load images and recreate them in different styles (which has given rise to a viral trend of recreating popular images in the style of studio Ghibli).
Who can access Images in ChatGPT? Upon its launch last week, the feature was only available to paid ChatGPT users. But as of yesterday…
OpenAI’s new image generator is now available to all users
In a post on X last night, Sam Altman announced that free ChatGPT users now have access to 4o image generation. As free users could previously create only three images per day with DALL-E 3, the expectation is that this same limit will hold for Images in ChatGPT.
Google unveils a next-gen family of AI reasoning models
Google has thrown its hat into the ring of competing reasoning models with the launch of Gemini 2.5, a family of reasoning models that includes Gemini 2.5 Pro Experimental, its most powerful model yet that has already risen to the top of the Chatbot Leaderboard. Like OpenAI, Google will incorporate reasoning capabilities into all of its models going forward.
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: Vibe Coding
Last week: Advanced ChatGPT
Lesson: Our class is a mix of digital media and computer science students. We spent a lot of time showing the computer science students how they use generative AI to create and edit media. Now we get to show the digital media students how they can use generative AI to write code.
That’s the idea behind vibe coding. These AI tools are so good at writing code that you can simply describe what you want and then ask it to troubleshoot any problems. No coding skills needed. We will talk more about this in Friday’s email newsletter, but to get started now use tools such as ChatGPT Canvas, Gemini Canvas and Claude Artifacts.
Your reading: The Vibe Coding Bible: How to Build Useful Things with Short Prompts, Speaking things into existence, Not a Coder? With A.I., Just Having an Idea Can Be Enough.
Homework: Our students will be put in groups to see what they can vibe code. So give it a try yourself and share what you create. Here’s a Tic Tac Toe game Professor Snider made in Cursor that is supposed to tell you the perfect move to win.
Tool of the week: Gemini Gems
Google Gemini users can now use its Gems feature for free. According to Gemini, “Gems are custom versions of Gemini that give you tailored responses. You can customize a premade Gem or you can create a new gem from scratch using instructions you set.” Gems are Gemini’s version of Custom GPTs. Try premade examples and create your own at gemini.google.com.
AI-generated image of the week
OpenAI’s new image creation tool is here, and suddenly we have a lot more control over images we create. Just look at all that text and attention to detail.
Prompt: two 40-year-old college professors standing in front of a college classroom. One has short gray hair and beard and glasses. He is holding up piece of cardboard with the words "Sign up for" written on it in hand lettering. The other is taller with short brown hair and brown beard. He is holding up a piece of cardboard with the words "our workshops" written on it in black marker. On a chalkboard behind them it says "April 9: AI in Law and Policy Lunch Club" and below that "April 11: Virtual Microsoft 365 Copilot Deep Dive " Add in any other details to make this look like an actual college class.What we found
Ethan Mollick wrote this week about image tools’ problem with elephants. Namely, if you used the prompt, “show me a room with no elephants in it, make sure to annotate the image to show me why there are no possible elephants", you would indeed get elephants.
I tried it in Midjourney this week:
Sure enough. Elephants. But this problem was overcome by OpenAI’s new 4o image model. I’ve had the same issue with trying to make images where someone was not wearing glasses. They always appear with glasses.