Hi everyone,
Welcome to all the new subscribers who signed up recently!
In this issue we’ll look at:
Meta’s new AI based on their Llama 3 model (try it at Meta.ai).
News of a new generative AI tool for music called Udio.
Several new AI devices that have been released or are coming soon (Humane AI Pin, Rabbit R1, Limitless Pin, and more).
Stories about AI in education, like Are We Asking the Wrong Questions About ChatGPT?
News about an AI Literacy Hackathon held at Carnegie Mellon University Libraries.
Several thought-provoking stories related to copyright, deepfakes, and more.
Links to two bots I made on Poe.
And much more…
Enjoy!
Foundation Model News
It’s worth trying Meta’s new AI. Access the web version at meta.ai. It’s free, open source, and can also create images – click on “Imagine” to try it. See their privacy guidelines: Privacy Matters: Meta’s Generative AI Features. You can also use it within Facebook Messenger, What’sApp, and Instagram. And you can use it without logging in (after verifying that you are over 13). It’s based on their model known as “Llama 3.”
Elon Musk’s ‘not bad’ review thrusts Meta’s Llama 3 AI into spotlight. VentureBeat.
”Released in open-source form today, Llama 3 represents a major leap forward for Meta’s AI ambitions. The model, which comes in 8 billion and 70 billion parameter versions, was trained on a staggering 15 trillion tokens of data, using a custom-built cluster of 24,000 GPUs. The result is a system that can go head-to-head with the best language models in the world, generating human-like text with uncanny fluency and coherence.”
OpenAI expands into Asia with Tokyo hub, Japanese variant of GPT-4. VentureBeat.
And they are releasing a custom version of GPT-4, which has been tuned specifically for the Japanese language
Google’s new technique gives LLMs infinite context. VentureBeat.
”A new paper by researchers at Google claims to give large language models (LLMs) the ability to work with text of infinite length. The paper introduces Infini-attention, a technique that configures language models in a way that extends their “context window” while keeping memory and compute requirements constant.”
Generative AI for multimedia
VASA-1: Lifelike Audio-Driven Talking Faces Generated in Real Time. Microsoft Research.
Take a look at the video examples. In addition to making the lips sync to the speech, it generates realist head movements, eyebrow movements and more. Microsoft says “we have no plans to release an online demo, API, product, additional implementation details, or any related offerings until we are certain that the technology will be used responsibly and in accordance with proper regulations.”
Adobe to add AI video generators Sora, Runway, Pika to Premiere Pro. VentureBeat.
”Adobe announced it aims to update Premiere Pro to add plug-ins to emerging third-party AI video generator models — including OpenAI’s Sora and rivals Runway ML’s Gen-2 and Pika 1.0.” It seems AI will be built into every tool!
Adobe launches Firefly 3 with full AI image generation in Photoshop. VentureBeat.
”… a fleet of Gen AI upgrades to Photoshop, including putting the image generation capabilities of its newly unveiled Firefly Image 3 AI model directly into the program. Adobe Firefly Image 3 will also exist as a stand-alone web app.”
Former Google DeepMind researchers launch AI-powered music creation app Udio. VentureBeat.
This one is similar to Suno (which I enjoy using). “The app allows users to create music by simply typing a description of the desired genre, providing lyrics or a subject, and indicating inspiring artists. Within 40 seconds, Udio generates a fully mastered, high-quality audio track. Users can then edit and iterate on their creations using the app’s “vary” feature, which enables them to adjust the sound and length of the track.”
What is Generative AI and Should Musicians be Afraid? Carlos Arana Says No! Berklee Online.
”Dr. Arana, who earned his PhD from University of Buenos Aires (UBA-UNLZ) in 2022, says generative AI is nothing for musicians to be afraid of. The only thing to be afraid of is ignorance.”
Generative AI devices
Several companies are experimenting with devices that have AI built-in. Here are some of the latest.
Humane Ai Pin: What went wrong and how it can be fixed (before it's too late). ZDNet.
This $700 AI pin that you wear on your shirt seemed to get bad reviews everywhere.
Rabbit R1 review: 24 hours with the AI device – unboxed and tested in hand. The Shortcut.
”I can confirm that this novel gadget is a fun yet unfinished device suitable for early adopters and AI chatbot enthusiasts willing to pay $199 to basically beta-test the technology.” Rabbit uses Perplexity for search.
(Here’s my saved search on Perplexity that tells you more about the Rabbit).
The Limitless Pendant.
Comparison review: Rabbit R1 vs Humane AI Pin vs Limitless Pendant: AI wearables compared. Tom’s Guide. Here’s the privacy policy from Limitless.
The Ray-Ban Meta Smart Glasses have multimodal AI now. The Verge.
”The primary command is to say “Hey Meta, look and...” You can fill out the rest with phrases like “Tell me what this plant is.” Or read a sign in a different language. Write Instagram captions. Identify and learn more about a monument or landmark. The glasses take a picture, the AI communes with the cloud, and an answer arrives in your ears.” “AI is a feature of the Meta glasses. It’s not the only feature. They’re a workable pair of livestreaming glasses and a good POV camera. They’re an excellent pair of open-ear headphones.”
Generative AI in education
Are We Asking the Wrong Questions About ChatGPT? Chronicle of Higher Education.
Many good points here.
Human and AI collaboration in the higher education environment: opportunities and concerns. Cognitive Research: Principles and Implications.
”Our fundamental recommendation is that higher education professionals can best serve learning outcome goals and the larger mission of education to provide a well-skilled workforce by embracing AI. There are clearly aspects of these new technologies that will make our jobs as educators more challenging. Nonetheless, we encourage the field to be creative and to see these tools as collaborators. AI can be a collaborator both for us, as instructors and educators, but also these systems must be accepted as collaborators of our students. Further, as students enter a workplace, their ability to work on human/computer “teams” is a critical cognitive skillset.”
Welcome to the Center for AI Learning. Emory University
They’ve opened a center for AI literacy, integration, and collaboration. I like this idea!
New AI coaching bot may enhance medical education, training. University of Arizona News.
Developing an AI bot to help medical students practice doctor/patient communication.
Generative AI in libraries
Using AI to Strengthen the Connection Between Technology and Scholarship. Against the Grain.
”The library is the conduit between information technology and content use, so as more universities embrace AI research and writing tools, libraries are becoming AI hubs where students and faculty can discover the best ways to use this evolving technology. “
Libraries AI Literacy Hackathon Expands Educational Resources. Carnegie Mellon University Libraries.
”The goal of the event was to gather academic librarians, staff, and other interested parties to develop open educational materials on the emerging principles of AI literacy.” “The hackathon had 36 attendees from across 17 different institutions, including The Ohio State University, West Virginia University, Princeton University, and Binghamton University.”
The Power of Retrieval Augmented Generation (RAG) LLM-based Academic Search Engines. Hong Kong University of Science & Technology.
”Recently, Mr. Aaron Tay, a librarian from Singapore Management University gave us a guest talk on RAG-based LLM applications for academic literature search. This post summarizes the key insights in the presentation.”
Thought-provoking
3 Myths About AI That Just Won't Die. Daniel Nest in Why Try AI?
I see a lot of people falling for Myth #3: “LLMs upgrade themselves on their own.” Worth reading.
AI chatbots refuse to produce ‘controversial’ output − why that’s a free speech problem. The Conversation.
Something to think about.
Creating structure with generative AI. Zach Seward’s talk at ISOJ 2024.
Some very cool use cases, with examples of creating summaries, extracting information, and structuring data in a more usable form — all with generative AI.
How One Author Pushed the Limits of AI Copyright. Wired.
Interesting and worth reading. “The USCO’s notice granting Shupe copyright registration of her book does not recognize her as author of the whole text as is conventional for written works. Instead she is considered the author of the ‘selection, coordination, and arrangement of text generated by artificial intelligence.’ This means no one can copy the book without permission, but the actual sentences and paragraphs themselves are not copyrighted and could theoretically be rearranged and republished as a different book.”
4 Shortcomings of Large Language Models - Yan LeCun, Research, and AGI. Synthedia.
Interesting thoughts on reasoning, planning, and memory.
Deepfakes, distrust and disinformation: Welcome to the AI election. Politico Europe.
”Many people’s political views are hard to budge, and experts warn that AI-generated content won’t change most voters’ party affiliations — no matter how compelling the made-up photo, video or audio clips appear.”
Just for fun
NBA Pacers used Snapchat AI filters to make it look like Los Angeles Lakers fans were crying during the game. Mario Nawfal on X.
Silly.
Robots Have Been About to Take All the Jobs for 100 Years. Pessimists Archive.
Sometimes it’s fun to look back.
Robort (A.I. Short Film). By Nobody and the Computer on YouTube.
Fun story of a guy who loses his job to AI, then disguises himself as a robot to get a job singing. (made with various AI image and video tools)
This camera trades pictures for AI poetry. TechCrunch.
A whimsical project. “People were fascinated by the concept of a camera that could generate poetry based on what it saw. The device quickly became the center of attention, sparking lively discussions and igniting the imagination of all who encountered it.”
Here’s a fun piece of AI-generated music from Udio (mentioned above).
Dune (as a Broadway musical).
My offerings
I’ve created two bots on Poe. Zoe Researcher helps undergrads choose library databases to search for their topic. Topic Chooser helps undergrads come up with a topic for a research paper. Let me know what you think.
Here you can see all of our offerings related to generative AI at University of Arizona Libraries. We have two Libguides, four tutorials, one assignment, and one FAQ. I’m working on an additional tutorial about generative AI for multimedia. We also have more FAQs in process that we hope to release in a few weeks. Everything is CC4.0, so feel free to copy and reuse.
The future
Why small language models are the next big thing in AI. VentureBeat.
”Enter small language models. SLMs are more streamlined versions of LLMs, with fewer parameters and simpler designs. They require less data and training time—think minutes or a few hours, as opposed to days for LLMs. This makes SLMs more efficient and straightforward to implement on-site or on smaller devices. One of the key advantages of SLMs is their suitability for specific applications. Because they have a more focused scope and require less data, they can be fine-tuned for particular domains or tasks more easily than large, general-purpose models.”
Move over, deep learning: Symbolica’s structured approach could transform AI. VentureBeat.
”At the heart of Symbolica’s approach is “category theory,” a branch of mathematics that formalizes mathematical structures and their relationships. By grounding AI in this rigorous framework, the company believes it can create models that have reasoning as an inherent capability, rather than an emergent side effect of training on huge datasets.”
Apple releases OpenELM: small, open source AI models designed to run on-device. VentureBeat.
”Just as Google, Samsung and Microsoft continue to push their efforts with generative AI on PCs and mobile devices, Apple is moving to join the party with OpenELM, a new family of open-source large language models (LLMs) that can run entirely on a single device rather than having to connect to cloud servers.”
Learn more
If you want to learn more about generative AI, contact me about doing a webinar for your group.
Later this year I’ll offer a 4-6 week asynchronous online course about generative AI. Sign up here if you’d like to be notified when it’s ready.
And as always, you can follow me on X, Mastodon, or Bluesky, where I post daily about generative AI.
And share this newsletter with your friends and colleagues!
What a fine newsletter issue, Nicole. So many good catches.
Hrm, now to figure out if I want to jump to iPhone for AI.