So far, we've mostly been focused on ways AI works with words (they are large LANGUAGE models after all). This week, we'll look at other generative AI capabilities, from pictures and video to music and narration. Let's create!
Subscribe to the BLAM! calendar and BLAM! announce list so you don't miss anything.
Reading:
Forbes, The Next AI Frontier: How Multimodal Systems Are Reshaping Our World
Google’s multi-modal FAQ and with examples prompts as well as how-to guides to help understand how to use Google’s multi-modal AI options
MIT Technology Review’s look at Google Astra is a great read once you watch the Google Project Astra Video
Video:
The Google Project Astra Video https://www.youtube.com/watch?v=nXVvvRhiGjI
Coursera's Generative AI for Everyone training from week 1 AI Basics video on Image Generation.
Try It: Use Gemini in Workspace or Dall-E in CBorg to generate a picture. Try one with text in the image - how does it do?
New! - Gemini for Google Workspace has been rolling out to all LBL Staff over the last week. If you don't have it yet, you will soon. Read about it here. We'll have a webinar later in BLAM! devoted just to these features.
New! Gemini uses Gemini 2.0 Flash across Google Workspace as of last week, so you should see faster results.
New! You can schedule an appointment to try out OpenAI's Deep Research with a Librarian or IT staff member. Try a complex research query and see how it does. Set up an appointment here.
OK, this is the one we've been waiting for - if you've noticed our header images they've all been leading up to this week's theme: Show us the Lab of the Future! Is it goats and researchers working together to advance science? Have the turkeys taken over the Lab? Are the deer-postdocs working late again? Robots everywhere? Cyborg PhDs? The future is up to you! You can create in any image creation program - gemini.google.com uses Imagen3 which is arguably the leader at the moment, but feel free to experiment with DallE in CBorg too. You might also want to try Google's ImageFX lab which gives Image3 more knobs and buttons.
Share your picture and your title/explanation for it in the BLAM Week 4 101 Chat Thread (Don't forget to read about how to use the BLAM Chat Rooms first). Or, if you'd like help with prompting, you can post your problem there too.
While you're there, peruse other people's submissions and upvote them by adding reactions. The BLAM! team will be awarding bucketlist awards to popular entries and our personal favorites. Oh, and keep it work-appropriate please.
NOTE: All past webinar recordings are available here.
Tuesday February 18th 2-3pm - For Coders, join Tim Fong as he walks you through how to create a multi file project with LBL Omni Engineer. Webinar recording here.
Wednesday February 19th Noon - 1PM- Webinar featuring a special guest speaker from Google, Goldy Arora, who will discuss all of the recent Google AI tools that are now available as part of our Google workspace including Gemini and much more. Webinar recording here.
Wednesday February 19th 11AM to Noon Coders Office Hours - Stop by and get answers to your questions about developing with LBL Omni Engineer.
Each Week, we'll feature some additional optional resources to dive deeper into AI.
This TechTarget article, What is Multi Modal AI, is a great overview of what is multimodal AI, how it is different from unimodal AI, and use cases.
MIT Technology Review's Multi-Modal AI Report goes deeper then our 101 content this week.
Ditto for IBM's intro to Multi-Modal, which includes some of the history and model differences.
Not Multi-Modal, but good thread for understanding RAG vs Context Window in actual performance.
CBorg has o1 mini and o3 mini available for testing. Try a difficult math, science, or multi-step analysis problem with them and see how it does.
New! You can schedule an appointment to try out OpenAI's Deep Research with a Librarian or IT staff member. Try a complex research query and see how it does. Set up an appointment here.
Can AI be your safety assistant (don't trust AI to be your actual safety assistant!)? Can you use multi-modal AI, either using photo inputs in CBorg or the Gemini app to assist with safety? Can you upload information about GERT, point at a door with a Radiation warning sign, and find out whether you can enter? Can AI review your ergonomics? Can AI help you plan work safely with the right inputs? Try it out. Again - don't use this for actual safety needs, this is just an exercise!
Ready to start integrating GenAI into your workflows? Google's cloudskillsboost site gives you several course options including "Integrate Generative AI Into Your Data Workflow" which starts with "Gemini for Data Scientists and Analysits".
OpenAI's o1 and o3 mini models are ranked among the highest for code development. Try giving them a difficult coding problem and see how they do. Try it on CBorg.
Multi-file programs are on the docket this week. Attend this week's webinar to learn how to use AI in larger projects.