Weekly Link Roundup #41

We'll get to all that in just a minute, but first a word about process....

So it’s been a minute since #40. I try to do this type of my newsletter, weekly at a minimum (hence the name) but sometimes it doesn’t work out that way and I was thinking about why. I think it’s because this isn’t just me reporting news or something - I think about the stories and look for connections - in my mind I see them as heat maps and as I thought about it, the answer was just as simple as the news the past week and half or so has been a mish mash and I just wasn’t seeing the patterns. Simple right? Ah, but simple doesn’t mean easy. I need to stay sharp looking for the patterns but also need to realize that sometimes they won’t be there. Anyway, here’s my fav quote about pattern recognition.
“We have no idea, now, of who or what the inhabitants of our future might be. In that sense, we have no future. Not in the sense that our grandparents had a future, or thought they did. Fully imagined cultural futures were the luxury of another day, one in which 'now' was of some greater duration. For us, of course, things can change so abruptly, so violently, so profoundly, that futures like our grandparents' have insufficient 'now' to stand on. We have no future because our present is too volatile. ... We have only risk management. The spinning of the given moment's scenarios. Pattern recognition” ― William Gibson, Pattern Recognition…..OK, back to this issue…

Definitely NOT me reading a prior newsletter. Courtesy of Hey Gen - and took all of 5 minutes to create.

Agents continuing to come to the fore as one of the main ways that individual level use of AI will happen.

Microsoft introduces autonomous AI agents: Here’s my question- does your org have any system or mechanism to programmatically review this capability? Why not? > > “In a blog post, Microsoft introduced ten autonomous agents for its Copilot AI platform, tailored to support sales, service, finance, and supply chain teams. These agents will be available for public preview starting in December, with a full rollout planned through 2025. Additionally, Microsoft announced the upcoming public preview of Copilot Studio, a tool enabling companies to create and manage their own AI agents.

Agent-as-a-Judge: Evaluate Agents with Agents: “Contemporary evaluation techniques are inadequate for agentic systems. These approaches either focus exclusively on final outcomes—ignoring the step-by-step nature of agentic systems, or require excessive manual labour. To address this, we introduce the Agent-as-a-Judge framework, wherein agentic systems are used to evaluate agentic systems.” I know - for me it is giving who watches the watchmen vibes but it could also improve and scale.

So I had this year in grad school when I thought there was a good chance I had ALS. Open and closed tube MRIs, CAT and PET scans, an EMG and even a muscle biopsy proved it was actually stress. I spent a year imaging though what it could be like. If this company can help patients with this interface, I’m so very here for it.

Unlocking the Potential of Brain-Computer Interfaces: “One person on our patient advisory board is Jules, who was diagnosed with ALS [Lou Gehrig’s disease] five years ago. People who have ALS, their brains continue to work totally cogently, but the brain’s ability to control the body deteriorates over time. The prognosis is three to five years of life expectancy.” Also this: A Neuralink Rival Says Its Eye Implant Restored Vision in Blind People.

After selling Anchor to Spotify, co-founders reunite to build AI educational startup Oboe: This company doesn’t even have a demo on their site - just a waiting list signup but man, am I interested. “The new company, hailing from Nir Zicherman and Michael Mignano, aims to democratize access to learning the way that their prior startup, Anchor, made it possible for anyone to create a podcast. That is, Oboe intends to produce a user-friendly interface that helps people accomplish the task at hand — in this case, expanding their knowledge via a combination of AI technology, audio, and video.” > > I want to know more. I want a white label platform that allows an org to unleash the creative power within it’s employees and even customers. I want a creator platform for L&D that can be deployed inside an enterprise.

A Million People Play This Video Wargame. So Does the Pentagon: Having worked on a crisis action planning ‘video game’ effort between OSD and Joint Staff J-7, I’m not stunned that commands are using commercial games for precise training - what amazes me is the stupidity of this quote from the article “Some officers have long derided wargames as entertainment, navel-gazing or “bogsats,” short for “bunch of guys standing around talking.” But the simulations—especially digital ones—can hone decision-making, spatial awareness and maneuvering abilities, say advocates.” I worked in the Pentagon, specifically on the non-kinetic training side for 5 years, for a few more years I was a consultant working with other govt agencies and then I was actually DoD civilian - worked on a number of games and game-based efforts and never once heard them run down like that.

Researchers say an AI-powered transcription tool used in hospitals invents things no one ever said: OK…this seems particularly bad: “Tech behemoth OpenAI has touted its artificial intelligence-powered transcription tool Whisper as having near “human level robustness and accuracy. But Whisper has a major flaw: It is prone to making up chunks of text or even entire sentences, according to interviews with more than a dozen software engineers, developers and academic researchers. Those experts said some of the invented text — known in the industry as hallucinations — can include racial commentary, violent rhetoric and even imagined medical treatments.”

From human experts to machines: An LLM supported approach to ontology and knowledge graph construction: “The conventional process of building Ontologies and Knowledge Graphs (KGs) heavily relies on human domain experts to define entities and relationship types, establish hierarchies, maintain relevance to the domain, fill the ABox (or populate with instances), and ensure data quality (including amongst others accuracy and completeness). On the other hand, Large Language Models (LLMs) have recently gained popularity for their ability to understand and generate human-like natural language, offering promising ways to automate aspects of this process. This work explores the (semi-)automatic construction of KGs facilitated by open-source LLMs.” > > Ross Dawson wrote a great summary of this paper here.

LinkedIn says it has verified 55 million users in effort to combat AI’s spread of scams, misinformation: You know that’s all cool and all but I’d also love LinkedIn to build a useful social graph of connections and a machine-readable standard for resumes.

File this under Unbelievably Stupid: Video game preservationists have lost a legal fight to study games remotely. The US Copyright Office needs to wake up and smell the new century “While the Copyright Office already lets institutions lend out other forms of media and even software programs remotely — so long as they don’t lend out more copies than they own — video games are still treated differently as of today.”

From Goodreads’ founder, Smashing debuts its AI-powered app for online readers: OK…file this under “I want to white label this and use it to create communities and interactions around content that I specify including enterprise content. “Like Goodreads, the app aims to create a community around content. But this time, instead of books, the focus is on web content — like news articles, blog posts, social media posts, podcasts, and more. In addition, Smashing is introducing an AI Questions feature that allows you to engage with the content being shared in different ways, including by viewing a news story from different perspectives or asking the AI to poke holes in the story, among other things.” Same with this: “Cafeteria raises $3M to connect brands with their most opinionated customers: teens.” And this “Anthropic’s AI can now run and write code”

Penguin Adds a Do-Not-Scrape-for-AI Page to Its Books: Something about a delayed closure of the barn door…..

The weird history of the barcode: I mean we look at em every day.