Weekly Link Roundup #45
45 - like the vinyl right? Kids, that's an "album"
Yes, the print is tiny. That's the point. There's something like 530+ agents listed here, doing everything from content creation, to marketing, to web development, and on and on. Now probably 90% of these will either fail or get acquired but the dynamic will change work. I not a 100% tech optimist (although I try and be positive), what I do know is that businesses, and rightly so, follow the money. If these agents can offer increases in efficiency, they will be adopted. Now would be a great time to get familiar with them, learn what they can and can't do and how to position yourself or your org accordingly. AI Agents Market Landscape - Ecosystem.
This is from LinkedIn’s 25 Big Ideas that will change our world in 2025. I usually don’t like year-end predictions because the groups making them almost never track how they did - no accountability. This is one I’ll include though because confirmation makes me warm and happy. I’ve been talking about creating models that let employees be creators and seems like someone is catching on. Why wouldn’t you create opportunities and tools for your own employees to serve as brand ambassadors or create L&D training content?
Ev Williams Was Lonely. He Doesn’t Want You to Be: I want to be excited about this but I’m not going to be. There are so many reasons why something like this is needed and as many ways that it can come off the rails and head down Enshittification Avenue. “On Thursday, they plan to unveil their app, Mozi, which is aimed at helping people foster in-person connections with their social circle. It lets people tell their friends about upcoming plans that may overlap. Those who join the app will see a private friend list based on their phone contacts.”
The new Surf browser shows why everyone’s trying to connect AI to the web: OK - not gonna lie - I think this is pretty cool. Read the article but this makes me think about a future of white label browsers or even personal-level browsers. I can see corporate laptops coming pre-installed with a browser that also contains all the info and data you need to do your job. As the new kid, when you need something, you just ask the chatbot. There’s a lot to unpack here in terms of potential. “But inside that demo is the big idea behind this browser, and a peek at why everyone’s so interested in connecting AI to the open web. Surf’s main character is the chatbot, which lives in the sidebar and has total access to everything you see and do in your browser. (Terrifying security nightmare? Maybe! Deta’s planning to do as much processing as possible locally, which should help.)”
The Algorithm and the Org Chart: How Algorithms Can Conflict with Organizational Structures: Anyone else looking forward to an algorithmic re-org? > > “Organizational structures, as often formalized into organization (org) charts, are meant to facilitate coordinated decision-making. Yet our 10-month ethnographic study of a large online retail company reveals why the organizational structures that facilitate effective decision-making by humans may be in tension with the organizational structures that facilitate effective decision-making using algorithms…This study suggests that as algorithms are integrated into the workplace, organization designs may begin to more explicitly reflect the contours of those algorithms' behaviors.”
The UC Berkeley Project That Is the AI Industry’s Obsession (gift link): Kinda surprised it took this long. Its “Hot Or Not” for AI > > Here’s the truth for 99% of most orgs - If “AI” development stopped today, and froze with current capabilities, it would still take 2-4 years for orgs, who were working hard, to implement and optimize org functions on current capabilities. “BERKELEY, Calif.—Record labels have the Billboard Hot 100. College football has its playoff rankings. Artificial intelligence has a website, run by two university students, called Chatbot Arena.”
AI Tutoring Outperforms Active Learning: “We find that students learn more than twice as much in less time when using an AI tutor, compared with the active learning class.”
The GPT Era Is Already Ending: A bad headline but a great (albeit long) read on the shift from predictive models to reasoning ones. If correct, potentially seismic.
This ex-Scale AI leader built a platform to automatically extracts insights from customer feedback: I wonder how most “learning content” would do with this level of feedback? How would L&D teams even handle it? “Customer interactions are a precious, yet immensely underutilized, dataset for all enterprises,” Varun told TechCrunch. “If they’re unlocked meaningfully, they can build best-in-class products and drive business growth.”
Generative AI is making traditional ways to measure business success obsolete: OK so I think the article is a little too breathless but its on the right track. If every L&D department right now doesn’t have at least one person or portions of multiple peoples’ days, looking at how to automate their low value activities, then I promise you, a vendor selling a service at a much lower price point than the L&D team, will soon visit your org’s VP. “Businesses operating a family of models known as "as a service" models, can make particular use of generative AI. In one of these, known as content-as-a-service (CAAS), firms provide other organizations with quick access to quality written content and visuals.”
Interesting Scientific Idea Generation Using Knowledge Graphs and LLMs: Evaluations with 100 Research Group Leaders: “Here, we introduce SciMuse, which uses 58 million research papers and a large-language model to generate research ideas. We conduct a large-scale evaluation in which over 100 research group leaders - from natural sciences to humanities - ranked more than 4,400 personalized ideas based on their interest. This data allows us to predict research interest using (1) supervised neural networks trained on human evaluations, and (2) unsupervised zero-shot ranking with large-language models. Our results demonstrate how future systems can help generating compelling research ideas and foster unforeseen interdisciplinary collaborations.”
Textbooks come alive with new interactive AI tool: How has it take this long and we don’t have this standard yet? “With just an iPad, students in any classroom across the world could soon reimagine the ordinary diagrams in any physics textbook—transforming these static images into 3D simulations that run, leap or spin across the page.”
AI Tutoring Outperforms Active Learning: “Here’s the big takeaway from this pre-print: We find that students learn more than twice as much in less time when using an AI tutor, compared with the active learning class.”
Character.AI has retrained its chatbots to stop chatting up teens: Filing this under “Too Little Too Late”
Sci-Fi Writer Arthur C. Clarke Predicts the Rise of Artificial Intelligence & Wonders What Will Happen to Humanity (1978): Well worth a watch. “Arthur C. Clarke, who notes that the dubious attitudes toward the prospect of thinking machines expressed in the late seventies had much in common with those about the prospect of space travel during his youth in the thirties. In his view, we were already “creating our successors. We have seen the first, crude beginnings of artificial intelligence,” and we would “one day be able to design systems that can go on improving themselves.”
AI-based tool creates simple interfaces for virtual and augmented reality: I feel like I should have this disclaimer ready in copy and paste - I post piece like this not because they are ready for prime time but to at least put them in the zeitgeist so we can start thinking about the implications now. “A paper published in Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology, by researchers in Carnegie Mellon University's Human-Computer Interaction Institute, introduces EgoTouch, a tool that uses artificial intelligence to control AR/VR interfaces by touching the skin with a finger.”