Weekly Link Roundup #38

This is a big one.....

OpenAI’s chief research officer has left following CTO Mira Murati’s exit: How many people can leave before OpenAI isn’t OpenAI anymore? “In Greek mythology, Theseus, the mythical king of the city of Athens, rescued the children of Athens from King Minos after slaying the minotaur and then escaped onto a ship going to Delos. Each year, the Athenians would commemorate this by taking the ship on a pilgrimage to Delos to honour Apollo. A question was raised by ancient philosophers: After several hundreds of years of maintenance, if each individual piece of the Ship of Theseus were replaced, one after the other, was it still the same ship?” See also > > OpenAI CTO Mira Murati is leaving.

MIT spinoff Liquid debuts non-transformer AI models and they’re already state-of-the-art: This caught my eye: “Unlike most others of the current generative AI wave, these models are not based around the transformer architecture outlined in the seminal 2017 paper “Attention Is All You Need.” Anything that is operating on a new architecture is interesting especially since this model seems to be more compact an efficient. This article also has more background on this “How MIT’s Liquid Neural Networks can solve AI problems from robotics to self-driving cars.” Now take these articles and go find the TPM in your org who is deep into AI, use them as a SME to build an understandable chunk of content on these developments including cost, security, and privacy implications and send that content to your senior leadership. That’s how #LearningAndDevelopment can quit being order takers.

Google NotebookLM leader says more controls coming for AI generated podcasts: Notebook is cool and the uses are fairly obvious - again, these kinds of developments cry out to be tried in order to be understood; “In a recent VentureBeat interview, Martin pointed out that corporate teams aInd educational users have increasingly turned to NotebookLM to streamline research and knowledge-sharing, suggesting that NotebookLM’s capabilities make it ideal for enterprises as well as individual users.” See also: Google's NotebookLM can now turn audio recordings and YouTube videos into interactive podcasts. Keep thinking that producing content the way you’ve always produced it is going to survive the next 24 months.

Y Combinator Traded Prestige for Growth: Have the mighty fallen? “Y Combinator is arguably the most successful early-stage VC fund or accelerator ever. The issue, however, is that they never truly grasped the factors behind their success, which is why YC's peak is already behind them—it’s likely all downhill from here.”

Instructors as Innovators: a Future-focused Approach to New AI Learning Opportunities, With Prompts: I get it. Why in the world would anyone in #LearningAndDevelopment and/or #Training want to read a paper by two professors one of whom researches learning and game design and one who is the preeminent researcher on AI experiments? “This paper explores how instructors can leverage generative AI to create personalized learning experiences for students that transform teaching and learning. We present a range of AI-based exercises that enable novel forms of practice and application including simulations, mentoring, coaching, and co-creation. For each type of exercise, we provide prompts that instructors can customize, along with guidance on classroom implementation, assessment, and risks to consider. We also provide blueprints, prompts that help instructors create their own original prompts.”

How a 15-Year-Old Gamer Became the Patron Saint of the Internet: THIS is deeply intriguing. Agree or disagree with the Catholic Church, the idea that we now have a techno-native saint is an interesting twist “After being named a "Servant of God" in 2013, Acutis reached the second rung on the ladder to sainthood when he was venerated by Pope Francis in 2018. His body was exhumed and brought to a tomb in Assisi where he still lies today, dressed in his trademark '90s teenager garb. “It’s a beautiful thing that for the first time in history you can see a saint dressed in jeans, sneakers, and a sweatshirt. That’s a great message.”

Massive E-Learning Platform Udemy Gave Teachers a Gen AI 'Opt-Out Window'. It's Already Over: Well that seems stupid and self-defeating - “Udemy, an e-learning platform with more than 250,000 online classes, recently announced that it would train generative AI on the classes that its users contribute to the site. Not only were class teachers automatically opted in to having their classes used as training, Udemy said teachers would have only a three-week “window” to opt-out of training. That window has now passed.” I do not know if this is in line with the TOS for Udemy but I assume it is - what gets me is why, after this, any instructor would willingly upload their course to Udemy.

The Internet Archive’s Fight to Save Itself: This institution needs to be protected.

California Passes Law Protecting Consumer Brain Data: Can someone call William Gibson and let him know we’re just living in his world now? “On Saturday, Governor Gavin Newsom of California signed a new law that aims to protect people’s brain data from being potentially misused by neurotechnology companies.”

AI startup Hugging Face reaches one million downloadable AI models – that’s a lot you have never heard of: “We just crossed 1,000,000 free public models on Hugging Face! “That’s the ones the media covers like Llama, Gemma, Phi, Flux, Mistral, Phi, Starcoder, Qwen, Stable diffusion, Grok, Whisper, Olmo, Command, Zephyr, OpenELM, Jamba, Yi but also 999,984 others. Why? “Because contrary to the “1 model to rule them all” fallacy, smaller specialized customized optimized models for your use-case, your domain, your language, your hardware, and generally your constraints are better.”

Science-Fiction Books Scientific American’s Staff Love: Did I mention that you should read SciFi in order to be better strategists and futurists? I think I did.

Discord opens Activities, in-app games and features, to all developers: Again - this happening on Discord is interesting but the model that it shows off is really interesting. How could you deploy a creator economy to help your org? “Discord announced today that it is opening development for Activities, its in-app ecosystem of apps and games, for all creators. This means that any developer can create, distribute and monetize apps on Discord via its Embedded Apps Software Development Kit. This means that Discord offers all developers an ecosystem on which to launch games and other software.”

What's in the White Space on your Organization Chart?: Such an important question that is almost NEVER considered. ONA can be a key unlock for an org.

Large models of what? Mistaking engineering achievements for human linguistic agency: “In this paper we argue that key, often sensational and misleading, claims regarding linguistic capabilities of Large Language Models (LLMs) are based on at least two unfounded assumptions: the assumption of language completeness and the assumption of data completeness. Language completeness assumes that a distinct and complete thing such as “a natural language” exists, the essential characteristics of which can be effectively and comprehensively modelled by an LLM. The assumption of data completeness relies on the belief that a language can be quantified and wholly captured by data.”

Plot Your Path: The 2024 AI Agent Ecosystem Map: Look, creating a map of anything that is changing as rapidly as the AI environment is just asking to be obsolete by the time you hit publish BUT Jeremiah has made his more useful and probably longer lasting by adding the layers. From an #innovation standpoint, the question will be, can we first decide which layers we need to test and then the kinds of experiments we need to conduct on each.

What to learn in the age of AI: Look, Gianni is wicked smart on AI and collective intelligence (#HUGEUnderstatement) and you should read the whole article but I want to highlight this one sentence “Don't try to outrun machines” - spot on! This is another way of saying my fav warning - don’t confuse your value with your activity. You can’t outrun the machines and you can’t out-activity them. You have to pick a different race to run. And hey, #LearningandDevelopment and #Training teams, this post is one you all need to deconstruct and use to plot a way out of 'order taking' and into providing higher order value to your orgs. Just sayin.

Sam Ladner’s Big Book Videos: Sam is the author of Practical Ethnography: A Guide to Doing Ethnography in the Private Sector and has been a principal researcher at both Amazon and Workday. Here she reviews some the books she considers to be important in this field.

Attention Is All You Need: Read the paper that introduced the transformer model of AI to the world.

Scenario Network Mapping by Dennis List, University of Adelaide, Australia: “This paper describes a new approach to scenario planning, known as scenario network mapping (SNM for short). The method, developed by the author in his doctoral thesis work, is contrasted with three more standard types of scenario planning. SNM differs from the conventional methods in using many more scenarios, each forming part of a particular pathway of possible events. In SNM, the focus is more on the network-like structure than on the scenarios themselves.”