Generative AI News - issue #9
Exploring AI’s potential and staying current with AI literacy
March 10, 2025
Hi everyone!
If you’re new here, welcome! I usually publish about once a month, but sometimes less often when my life gets busy (like recently).
A lot has been happening in the world of genAI, so let’s jump right into it.
In this issue we’ll look at:
News about ChatGPT 4.5, Claude 3.7, & some interesting updates in Google’s AI Studio.
“DeepResearch” is available in so many places.. there’s a lot to experiment with here — and what does it mean for the future of library research?
Developing AI Literacy: Reflections on Teaching a Six-week Course for Library Workers - my recent blog post on LibTech Insights.
Several interesting stories about genAI for improving accessibility.
What’s happening with genAI in education and in libraries.
Many thought-provoking articles, as usual! (like this one about copyright and this one about AI and climate)
Enjoy!
Foundation Models
OpenAI
OpenAI expands GPT-4.5 rollout. Here's how to access (and what it can do for you) - Sabrina Ortiz, ZDNet
”On Wednesday morning, OpenAI announced via an X post that it began rolling out GPT-4.5 to ChatGPT Plus users. When first announced, OpenAI shared that the full rollout could take one to three hours. However, just an hour later, the full rollout of GPT-4.5 was completed, which was faster than expected, according to the X post. “
OpenAI releases ‘largest, most knowledgable’ model GPT-4.5 with reduced hallucinations and high API price - Carl Franzen, VentureBeat
ChatGPT update adds canvas sharing and improved reasoning transparency - Alexey Shabanov, TestingCatalog
”Users can now share their canvases by generating a public URL. When someone opens this link, they will see a rendered version of the canvas. …This feature is similar to Artifact Remix on Claude, enabling easier collaboration and content sharing.”
Too many models, too much confusion: OpenAI pledges to simplify its product line - Emilia David, VentureBeat
”OpenAI’s decision to simplify how it presents models points to a continuing problem: Not everyone understands what each model is supposed to do and what makes it different. OpenAI is not alone in this.”
Navigating the confusing names of AI models (13 page PDF) - Tianyu Xu on LinkedIn
Here’s a direct link to the PDF. Hopefully this won’t be needed if/when OpenAI simplifies things.
Anthropic
Claude 3.7 Sonnet: Fantastic Model Held Back by Lack of Native Internet Access, Daniel Nest, Why Try AI
(I heard the CEU of Anthropic say they will add web search later this year).
Anthropic’s Claude 3.7 Sonnet takes aim at OpenAI and DeepSeek in AI’s next big battle - Michael Nuñez, VentureBeat
Apple
Apple prepares to add Google Gemini to Apple Intelligence - William Gallagher, Apple Insider
Microsoft
Announcing Free, Unlimited Access to Think Deeper and Voice - Microsoft
Microsoft’s new Phi-4 AI models pack big performance in small packages - Michale Nuñez, VentureBeat
Google
Enhancing video navigation and accessibility with Google Drive video transcripts - Google Workspace Updates
Google Search’s new ‘AI Mode’ lets users ask complex, multi-part questions - Aisha Malik, TechCrunch
”AI Mode is rolling out to Google One AI Premium subscribers starting this week and is accessible via Search Labs, Google’s experimental arm.” (video demo)
Reference past chats for more tailored help with Gemini Advanced - Dave Citron, Google
”Starting today, Gemini can now recall your past chats to provide more helpful responses. Whether you’re asking a question about something you’ve already discussed, or asking Gemini to summarize a previous conversation, Gemini now uses information from relevant chats to craft a response. That means no more starting over from scratch or having to search for a previous conversation thread.”
You knew it was coming: Google begins testing AI-only search results - Ryan Whitwam, ArsTechnica
”Everyone will soon see more AI Overviews at the top of the results page, but Google is also testing a more substantial change in the form of AI Mode. This version of Google won't show you the 10 blue links at all—Gemini completely takes over the results in AI Mode.”
Data Science Agent in Colab: The future of data analysis with Gemini - Fine et al., Google for Developers
DeepSeek
DeepSeek rushes to launch new AI model as China goes all in - Eduardo Baptista, Julie Zhu and Fanny Potkin, Reuters
DeepSeek to share some AI model code, doubling down on open source - Reuters
DeepSeek and shallow moats: what does it mean for higher education? - Ben Swift, Times Higher Education
”Despite trillions of dollars of investment, it really does still seem like an upstart can come out of nowhere to release – and share – something that’s competitive with state-of-the-art offerings from the tech giants. This poses a strategic question for higher education leaders: how should institutions position themselves in response?”
Other models
Mistral releases new optical character recognition (OCR) API claiming top performance globally - Carl Franzen, VentureBeat
”Unlike traditional OCR solutions that primarily focus on text extraction, Mistral OCR is designed to interpret various document typographical elements and characters, including tables, mathematical expressions and interleaved images, while maintaining structured outputs.”
Alibaba’s new open source model QwQ-32B matches DeepSeek-R1 with way smaller compute requirements - Carl Franzen, VentureBeat
”… it’s available for commercial and research uses, so enterprises can employ it immediately to power their products and applications (even ones they charge customers to use). It can also be accessed for individual users via Qwen Chat.”
Introducing Poe Apps - Poe.com
”Today we are introducing Poe Apps, which make it easy to build visual interfaces on top of any combination of the existing models on Poe and custom logic expressed in JavaScript.”
Manus AI agent could be the next "Deepseek moment" for Western AI labs - Jonathan Kemper, The Decoder
Use cases on the Manus website. Waiting list for now. Brent Anders discusses what it means for academia.
Images, video, music, and voices
Video
From the YouTube CEO: Our big bets for 2025 - Neal Mohan, YouTube
”As we look ahead, we'll continue investing in AI tools that empower creators and artists throughout their creative journey. Last year we rolled out Dream Screen and Dream Track which generate image backgrounds, video backgrounds and instrumental soundtracks for Shorts. We’ll continue investing in these features, including integrating Veo 2 into Dream Screen soon.”
ChatGPT app could soon generate AI videos with Sora - Best of AI
”OpenAI is planning to integrate its Sora text-to-video generation tool into the ChatGPT app, expanding its availability beyond the current web client exclusive to paid users.”
Alibaba Makes AI Video Generator Wan 2.1 Free to Use - Pesala Bandara, PetaPixel
”Alibaba announced that four models that are part of its Wan 2.1 series are now open source and can be downloaded and modified by users. Wan 2.1 can generate images and video from text and image inputs.” (See examples in this review from Curious Refuge on YouTube).
Voice
Perplexity's voice mode gets a futuristic makeover on your iPhone - Eric Hal Schwartz, Tech Radar
Octave TTS: the first text-to-speech system that understands what it’s saying - Hume
Hume launches new text-to-speech model Octave that generates custom AI voices with adjustable emotions - Carl Franzen, VentureBeat
ElevenLabs’ new speech-to-text model Scribe is here with highest accuracy rate so far (96.7% for English) - Carl Franzen, VentureBeat
Alexa+ is Amazon's AI answer to... well, pretty much everything - The Neuron
”…it uses both Amazon's Nova and Anthropic's Claude models, selecting the best AI for any specific task.” Free for Prime members or $19.99/month, officially rolls out in the US over the next few weeks via an early access period.
Rebuilding Alexa: How Amazon is mixing models, agents and browser-use for smarter AI - Emilia David, VentureBeat
Crossing the uncanny valley of conversational voice - Brendan Iribe, Ankit Kumar, and the Sesame team
(Try their demos).
Music
NotaGen: Advancing Musicality in Symbolic Music Generation with Large Language Model Training Paradigms - Wang et al.
Includes several classical music examples on the bottom of the page.
DeepResearch in all its forms
These experts were stunned by OpenAI Deep Research - Timothy B. Lee, Understanding AI
OpenAI expands Deep Research access to Plus users, heating up AI agent wars with DeepSeek and Claude - Michael Nuñez, VentureBeat
A Note on Deep Research Products and the Return of the Traditional Credibility Chain - Mike Caulfield, The Ends of Argument
The Deep Research problem - Benedict Evans
Interesting commentary.
Perplexity just made AI research crazy cheap—what that means for the industry - Michael Nuñez, VentureBeat (on pricing for Deep Research)
”The launch exposes a painful truth in AI pricing: Expensive enterprise subscriptions may be unnecessary. While Anthropic and OpenAI charge thousands monthly for their services, Perplexity offers five free queries daily to all users. Pro subscribers pay $20 monthly for 500 daily queries and faster processing — a price point that could force larger AI companies to explain why their services cost up to 100 times more.”
How we evaluated Elicit Reports - Étienne Fortier-Dubois, Elicit
”We recently announced Elicit Reports: fully-automated research overviews for actual researchers, inspired by systematic reviews.”
A Practical Guide to Implementing DeepSearch/DeepResearch - Han Xiao, Jina AI
”In this article, we'll discuss the principles of DeepSearch and DeepResearch by looking into our open-source implementation.” (very interesting)
Tips
Prompt Engineering Guide for Deep Research with ChatGPT’s O3-Model - Max van den Broek on LinkedIn
Writing effective text prompts for video generation - Adobe Firefly
Here Are My Go-To AI Tools - Daniel Nest, Why Try AI
A useful list of his preferred LLMs, image models, AI video sites, music makers, and research tools.
Open AI: compare models - OpenAI
You can now compare their models in a handy chart.
How to master new topics with AI - Daniel Nest, Why Try AI
Ideas for prompts like this: ““Summarize key themes and insights in the attached sources. Identify any inconsistencies or knowledge areas that aren’t sufficiently covered. Provide potential research questions that might help address these knowledge gaps.”
Private, Useful, and Optional AI: DuckDuckGo offers free access to popular AI chatbots at Duck.ai and expands AI-assisted answers - DuckDuckGo News
Good for privacy: “Get free, anonymized access to popular chatbots at Duck.ai.”
My offerings
Developing AI Literacy: Reflections on Teaching a Six-week Course for Library Workers - Nicole Hennig, LibTech Insights
My reflections on teaching the course in November/December 2024. I’m offering it again via Infopeople, beginning April 15. Learn more about the course or register now.
ACRL AI Competencies for Academic Library Workers - PDF, draft ready for comment
I’ve been on this task force for several months. Now we have a draft ready for comments from library workers. Here’s the feedback form (fill out by Mar. 26).
The Creative Resister - custom GPT and The Creative Resister bot on Poe - both by Nicole Hennig
Here are my bots for websites facing pressure to remove DEI or climate-change related content. They are based on ideas from Timothy Snyder's book On Tyranny. (Not meant to suggest obeying in advance). Here’s a sample output from Poe you can read without a Poe account.
Generative AI Sparks New Library Frontiers - episode 2 of The Infinite Stream by Kyle Bylin
I was interviewed for this episode of The Infinite Stream Podcast.
Accessibility
Meta’s AI-Powered Ray-Bans Are Life-Enhancing for the Blind - Sarah E. Needleman, Wall St Journal
”After Allison Pomeroy lost most of her vision two years ago, her husband began reading menus, signage and other text out loud to her. He doesn’t need to anymore".”
Nvidia helps launch AI platform for teaching American Sign Language - Dean Takahashi, GamesBeat
”Nvidia, the American Society for Deaf Children and creative agency Hello Monday are helping close this gap with Signs, an interactive web platform built to support ASL learning and the development of accessible AI applications.”
AI Audiobooks & Learning - Lance Eaton, AI + Education = Simplified
”Of course, there is an argument that AI will take jobs from artists like narrators. That is, companies and self-publishers will opt to use an AI narrator and therefore, there will be fewer paying opportunities and those will inevitably go to the more established and popular narrators (the rich get richer). And there’s truth in that, but in the context of the world we live in, it also becomes an anti-accessibility argument. There are amazing entities like the Learning Ally (formerly Recording for the Blind & Dyslexic) and National Library Service for the Blind and Print Disabled doing important work to make books accessible to those with visual disabilities and yet, they are never going to be able to keep up with the millions of books published each year, never mind the backlog of millions of books that have never been converted to audio.”
Notes from an Accessibility and Gen AI podcast appearance - Simon Willison
”Some people I’ve talked to have been skeptical about the accessibility benefits because their argument is that if you give somebody unreliable technology that might hallucinate and make things up, surely that’s harming them. I don’t think that’s true. I feel like people who use screen readers are used to unreliable technology… When you consider that people with accessibility needs have agency, they can understand the limitations of the technology they’re using. I feel like giving them a tool where they can point their phone at something and it can describe it to them is a world away from accessibility technology just three or four years ago.
What’s happening in education
Creativity with AI: New Report Imagines the Future of Student Success, Brian Johnsrud, Adobe Blog
Building an AI-Ready Workforce: A Look at College Student ChatGPT Adoption in the US - OpenAI. (PDF)
Includes top student use cases from OpenAI user data: learning/tutoring, writing help, misc. questions, and programming help.
Estonia provides free ChatGPT for its entire secondary education system - Stefan Bauschard, Education Disrupted
Designing courses with an AI assistant - Jeremy Caplan and Maime Rheingold, ChatGPT for Education
Shaping integrity: why generative artificial intelligence does not have to undermine education - Myles Joshua Toledo Tan and Nicholle Mae Amor Tan Maravilla, Front. Artif. Intell. , 23 October 2024
”This discussion will explore how GAI can be integrated into education in ways that support rather than erode academic integrity. By examining the ethical frameworks of deontological ethics and consequentialism, and educational theories like constructivist learning and SDT, we will argue that AI, when used responsibly, can enhance digital literacy, foster intrinsic motivation, and support genuine knowledge construction.”
Audio-Based Learning 4.0 - Dr. Philippa Hardman
”What's super interesting is how the solid research backing audio's effectiveness is and how well this is converging with these new AI capabilities. Tools like ElevenLabs and Google MusicFX are helping to remove a lot of barriers that have made creating quality learning audio content challenging. It’s now easier than ever for anyone interested in innovating their pedagogical approaches to start experimenting with evidence-based audio content in their work.”
‘Generative AI is making our students more creative than ever’ - Ramona Pistol, Times Higher Education
”The future of education isn’t about AI creating content – it’s about students becoming more creative through their interaction with AI. It’s time for our teaching practices to reflect this reality. One way to do this is to embrace GenAI as a collaborative partner and facilitate GenAI-enhanced project-based learning by designing projects relevant to real-world issues that allow students to use it creatively.”
How is AI Changing What it Means to Learn? - The Centaur by AI Consensus
Thoughts and examples from grad students.
AI legibility, physical archives, and the future of research - Benjamin Breen, ResObscura
”In other words: physical research and critical thinking about the nature of archives and qualitative data just got way more important. Not less.”
Introducing NextGenAI: A consortium to advance research and education with AI - OpenAI
”NextGenAI’s founding partners are Caltech, the California State University system, Duke University, the University of Georgia, Harvard University, Howard University, Massachusetts Institute of Technology, the University of Michigan, the University of Mississippi, The Ohio State University, the University of Oxford, Sciences Po, Texas A&M University, as well as Boston Children’s Hospital, the Boston Public Library, and OpenAI.”
Examples:
“Texas A&M is using NextGenAI resources to fuel their Generative AI Literacy Initiative, providing hands-on training to enhance the responsible use of AI in academic settings.”
“MIT students and faculty will be able to use OpenAI’s API and compute funding to train and fine-tune their own AI models and develop new applications.”
“Howard will use AI to develop curricula, experiment with new teaching methods, improve university operations, and give students hands-on AI experience to prepare them as future leaders.”
What’s happening in libraries
Outlook on Federal Funding Threats to Libraries: What library workers need to know to advocate for their institutions - Larra Clark, American
Not about AI, but important to know about.
Libraries supporting data rescue - Data Rescue Project
Also not about AI, but important to know about.
AI Literacy: A Guide for Academic Libraries - Leo S. Lo, College & Research Libraries News
”Cross-cutting themes.” Human-AI collaboration, lifelong learning & adaptation, equity & access.
From Here to AI Transformation - Carlo Iacono, Hybrid Horizons
”The standard committee structure with representation from all stakeholders and careful deliberation over time serves many purposes well. But it falters in the face of rapid technological change. Establishing small, interdisciplinary response teams with the authority to implement changes without navigating the full bureaucratic labyrinth become necessary. These teams combine technical expertise, pedagogical wisdom, and operational authority.”
Cutting through the noise: Assessing tools that employ artificial intelligence - Norgueria et al., IFLA Journal
See tables 1 and 2 for “guiding questions for reflection when assessing innovative AI tools and services.”
Artificial Intelligence in PhD education: New perspectives for research libraries - Grote et al., Liber Quarterly
”The study reveals a degree of uncertainty among librarians about their role in the AI academic nexus. For the development of competences of teaching staff in academic libraries, the paper recommends to integrate AI-related topics into existing educational resources and to create arenas for sharing experiences and knowledge with relevant partners both within and outside the university.”
Just for fun
Nobody Believes Cassandra (or, The Tragedy of Cassandra in the Age of Artificial Intelligence) - James Maynard, AI Creator House on YouTube
“What if the key to saving the future lies in the past? 🤔 In Nobody Believes Cassandra, Dr. Cassandra Selene creates HELEN, an AI designed to decode history—but when it starts predicting the future, everything changes. 🌍⏳”
Take a look at this 54-minute film, made with numerous AI tools by my husband, James. It took him several months to create. Check out his website, AI Creator House, for more films he’s made with the help of AI.
Thought-provoking
New research & techniques
Inception emerges from stealth with a new type of AI model - Marina Temkin, TechCrunch
”Inception, a new Palo Alto-based company started by Stanford computer science professor Stefano Ermon, claims to have developed a novel AI model based on “diffusion” technology. Inception calls it a diffusion-based large language model, or a “DLM” for short.” (faster with reduced computing costs)
OctoTools: Stanford’s open-source framework optimizes LLM reasoning through modular tool orchestration - Ben Dickson, VentureBeat
”OctoTools, a new open-source agentic platform released by scientists at Stanford University, can turbocharge large language models (LLMs) for reasoning tasks by breaking down tasks into subunits and enhancing the models with tools.”
A new generation of AIs: Claude 3.7 and Grok 3 - Ethan Mollick
”Together, these two trends are supercharging AI abilities, and also adding others. If you have a large, smart AI model, that can be used to create smaller, faster, cheaper models that are still quite smart, if not as much as their parent. And if you add Reasoner capabilities to even small models, they get even smarter. What that means is that AI abilities are getting better even as costs are dropping.”
Well, it looks like Meta's Yann LeCun may have been right about AI - again - Maximillian Schreiner, The Decoder
“A new study led by Meta's Head of AI Yann LeCun demonstrates how artificial intelligence can develop basic physics understanding just by watching videos. The findings support LeCun's alternative vision to generative AI and challenge approaches like OpenAI's Sora.”
Researchers find you don’t need a ton of data to train LLMs for reasoning tasks - Ben Dickson, VentureBeat
”Their findings show that with just a small batch of well-curated examples, you can train an LLM for tasks that were thought to require tens of thousands of training instances. This efficiency is due to the inherent knowledge that modern LLMs obtain during the pre-training phase. With new training methods becoming more data- and compute-efficient, enterprises might be able to create customized models without requiring access to the resources of large AI labs.”
Mayo Clinic’s secret weapon against AI hallucinations: Reverse RAG in action - Taryn Plumb, VentureBeat
”The hospital has employed what is essentially backwards RAG, where the model extracts relevant information, then links every data point back to its original source content. Remarkably, this has eliminated nearly all data-retrieval-based hallucinations in non-diagnostic use cases — allowing Mayo to push the model out across its clinical practice.”
Chain of Draft: Thinking Faster by Writing Less - Xu et al., arxiv.org
”In this work, we propose Chain of Draft (CoD), a novel paradigm inspired by human cognitive processes, where LLMs generate minimalistic yet informative intermediate reasoning outputs while solving tasks. By reducing verbosity and focusing on critical insights, CoD matches or surpasses CoT in accuracy while using as little as only 7.6% of the tokens, significantly reducing cost and latency across various reasoning tasks.”
Surveys
Goodbye Google? People are increasingly switching to the likes of ChatGPT, according to major survey – here’s why - Chris Rowlands, TechRadar
”Survey data suggests chatbots are replacing search engines.”
National AI opinion monitor: AI trust and knowledge in Ameria - Katherine Ognyanova and Vivek Singh, Rutgers University (PDF)
”Here, we examine public trust in AI, in the companies that use it, and in the news content produced by it. We also evaluate people’s objective and perceived knowledge about artificial intelligence.”
Copyright
AI and Copyright: Expanding Copyright Hurts Everyone—Here’s What to Do Instead - Tori Noble, Electronic Frontier Foundation
”You shouldn't need a permission slip to read a webpage–whether you do it with your own eyes, or use software to help. AI is a category of general-purpose tools with myriad beneficial uses. Requiring developers to license the materials needed to create this technology threatens the development of more innovative and inclusive AI models, as well as important uses of AI as a tool for expression and scientific research.”
Breaking down a federal court's ruling on AI and copyright - Mathew Ingram, The Torment Nexus
”I saw headlines that claimed this was a "landmark ruling" and a "major win" for content creators, that the AI industry was on the ropes, that this decision marked a turning point in the debate over copyright and artificial intelligence, that fair use is over as a defense for AI training, etc. etc. Is this accurate? Not really. It is definitely true that the court's ruling is the first significant federal decision related to AI and copyright. But there are a number of reasons why this case doesn't have as much impact on AI and copyright as the headlines might lead you to believe.”
Open source
Defending Open Source AI Against the Monopolist, the Jingoist, the Doomer and the Idiot - Daniel Jeffries, Future History
”Whenever you hear something like “so and so is expert in AI and they say AI will take over the world,’ the first question to ask yourself here is "expert in what? Just because you are an expert in the inner workings of AI, does not mean you are an expert in psychology, business, economics, history, or how technology develops and changes over time.”
Climate
How much energy does ChatGPT use? - Josh You, Epoch AI
”A commonly-cited claim is that powering an individual ChatGPT query requires around 3 watt-hours of electricity, or 10 times as much as a Google search.1 This is often brought up to express concern over AI’s impact on the environment, climate change, or the electric grid. However, we believe that this figure of 3 watt-hours per query is likely an overestimate. In this issue, we revisit this question using a similar methodology, but using up-to-date facts and clearer assumptions. We find that typical ChatGPT queries using GPT-4o likely consume roughly 0.3 watt-hours, which is ten times less than the older estimate. This difference comes from more efficient models and hardware compared to early 2023, and an overly pessimistic estimate of token counts in the original estimate.”
Here’s How We Can Power the AI Boom Without Building a Ton of New Gas Plants - Alexander C. Kaufman, Mother Jones
”… new research from Duke University shows how we could do four Stargate Projects without building a single new power plant. The study, published Tuesday morning, found that scaling back power usage by just 0.5 percent per year could free up supply on the grid for nearly 100 gigawatts of additional power demand.”
Microsoft powers AI ambitions with 400 MW solar purchase - Tim De Chant, TechCrunch
”Microsoft has added another 389 megawatts of renewable power to its portfolio as the tech giant scrambles to meet the power demands required to match its AI ambitions. The additional renewable power spans three solar projects developed by EDP Renewables North America — two in southern Illinois and one outside Austin, Texas.”
Deepfakes
Deepfakes and the 2024 US Election - Bruce Schneir, Schneir on Security
” ‘We find that (1) half of AI use isn’t deceptive, (2) deceptive content produced using AI is nevertheless cheap to replicate without AI, and (3) focusing on the demand for misinformation rather than the supply is a much more effective way to diagnose problems and identify interventions.’ “
“This tracks with my analysis. People share as a form of social signaling. I send you a meme/article/clipping/photo to show that we are on the same team.”
Explanations
Scaling up: how increasing inputs has made artificial intelligence more capable - Veronika Samborska, Our World in Data
”The path to recent advanced AI systems has been more about building larger systems than making scientific breakthroughs.” (useful, simple explanations)
Demystifying Reasoning Models - Cameron R. Wolfe, Ph.D., Deep (Learning) Focus
”Recently, a completely new paradigm in LLM research has emerged: reasoning. Reasoning models approach problem solving in a completely different manner compared to standard LLMs. In particular, they spend a variable amount of time “thinking” prior to providing their final answer to a question.” (Long and interesting technical explanation).
Inspiration
Three Fallacies: Alondra Nelson's Remarks at the Elysée Palace on the Occasion of the AI Action Summit - Alondra Nelson, Tech Policy Press
”We can create systems that expand opportunity rather than concentrate power. We can build technology that strengthens democracy rather than undermines it. Now, as we shape what may be humanity’s most transformative innovation, we must ensure AI follows this tradition.”
Beneficial uses of AI
How AI can help fact checkers - Dr. David Corney, Full Fact
Announcing the winners of the Frugal AI Challenge 🌱 - Sasha Luccioni and Theo Alves, Hugging Face
An AI model that can decode and design living organisms - Andrew Maynard, The Future of Being Human
AI cracks superbug problem in two days that took scientists years - Tom Gerken, BBC
Google’s new AI co-scientist aims to speed up the scientific discovery process - Gyana Swain, ComputerWorld
Face readers: Artificial intelligence is becoming better than humans at scanning animals’ faces for signs of stress and pain - Christa Lesté-Lasserre, Science
”As each pig squeezes into her feeding stall, a tiny camera snaps a photo of her face.” (I like the name: “Intellipig” - good for animal welfare)
Mistral Saba - Mistral AI
”Mistral Saba is a 24B parameter model trained on meticulously curated datasets from across the Middle East and South Asia. The model provides more accurate and relevant responses than models that are over 5 times its size, while being significantly faster and lower cost. The model can also serve as a strong base to train highly specific regional adaptations.”
ChatGPT Saved My Life (No, Seriously, I’m Writing this from the ER) - Bethany Crystal
”How using AI as a bridge when doctors aren't available can improve patient-to-doctor communications in real time emergencies.”
Cohere’s first vision model Aya Vision is here with broad, multilingual understanding and open weights - Carl Franzen, VentureBeat
”Aya Vision’s capabilities have broad implications across multiple fields:
• Language learning and education: Users can translate and describe images in multiple languages, making educational content more accessible.
• Cultural preservation: The model can generate detailed descriptions of art, landmarks and historical artifacts, supporting cultural documentation in underrepresented languages.
• Accessibility tools: Vision-based AI can assist visually impaired users by providing detailed image descriptions in their native language.”
The future
What AI can currently do is not the story - Ege Erdil, Epoch AI
”If you want to predict the future of AI beyond a horizon of a few years, it’s not a good idea to anchor on the present capabilities of AI systems, or even on the present rate of change of those capabilities. You should instead think in terms of the intrinsic difficulty of various tasks, many of which current AI systems cannot perform at all, and what that means for the resource demands of automating those tasks using AI.”
Hugging Face co-founder Thomas Wolf just challenged Anthropic CEO’s vision for AI’s future - Michael Nuñuz, VentureBeat
”Wolf’s critique doesn’t suggest abandoning current approaches, but rather augmenting them with new techniques and metrics specifically aimed at fostering contrarian thinking. In his post, Wolf suggests that new benchmarks should be developed to test whether scientific AI models can “challenge their own training data knowledge” and “take bold counterfactual approaches.””
The Model is the Product - Vintage Data
”To reassert: the big labs are not advancing with an hidden agenda. While they can be opaque at time, they laying it all in the open: they will bundle, they will go up the application layer and they will attempt to capture most of the value there.”
Learn more
My six-week online course, AI Literacy for Library Workers, is happening again via Infopeople, beginning April 15. The Nov/Dec session went well and I enjoyed teaching it! You can register here.
Here you can read some comments from people who participated in the Nov/Dec session of the course.
thanks - love your newsletter! It's great to have a library specific focus on AI (as well as reading some of the general stuff)
Another fantastic issue, thanks for all the new tabs I have to deal with! :-)