Weekly Link Roundup #63
Soooooo....kind of a big week....
I’ve got a few items here that don’t fit into the standard format for this newsletter but I checked with the Editor-in-chief (that’s me) and he agreed with the head writer (also me), that we could change things up a little for this issue. Just to be sure, we all ran it by our Ombudsman (ironically, me), and he also agreed.
Item #1: I started a new job - on Cinco de Mayo no less! I’m now the Director of Customer Education (Basecamp) at Qualtrics. I’m super excited to jump into this role - I think it’s more important than ever to be able to listen to both your customers and employees and Qualtrics is at the leading edge of that. I wrote more about it here - and thanked a ton of people - keep building that network kids! For now though, just know that I’m psyched, pumped, and ready to rock!
Item #2: Josh Cavalier writes a great post - OK, that on its own, is not news, Josh writes a lot of great posts. This one though was his recap of what he saw coming out of the ServiceNow Knowledge 2025 conference. If you work in #LearningAndDevelopment and don’t go read this post, you’re going to be real surprised (and probably not in a good way) when certain things start to happen. Let me add some context - do you all remember when L&D leaders were consulted about whether or not we liked or wanted Sharepoint? Nope, you don’t. It just showed up one day because it came with the MSFT enterprise license.
Well what ServiceNow announced (see Josh’s post) is poised to have major impacts across the L&D world. Even if your company doesn’t use ServiceNow, this offering will shape other companies’ offerings. If L&D teams don’t stop confusing their value with their activity and find ways to convince sr leadership that they need to be in the room for these decisions, they will be stuck with a system they didn’t choose and evaluated and rated on how well they implement it.
This isn’t like which LMS to pick - these are enterprise-wide systems that will suck up the learning function like an afterthought if you’re not in the room to shape that decision. Get on it.
Item #3: Had a great Institute for Innovation in Large Organizations (ILO) call - OK, again, like Josh’s posts, ILO calls are usually great but this one was focused on what changed during COVID that has stuck (and a little on what hasn’t). One thing the continues to hit me is the divide between the companies (like Qualtrics and Dropbox), who have done the work and who have ben thoughtful about being either remote-first or hybrid-first and those that haven’t.
We hear a lot from those that haven’t about not wanting to miss that serendipity of being in the office together, those magical, accidental hallway moments, those golden nuggets of genius that spring from deskside drive-bys - ah, the golden days of 5 days in and no questions. Here’s the thing. No one put any thought into creating those opportunities - they just happened because we were in physical proximity. You know what you didn’t see? You didn’t see the connections that get made and the ideas that get created when your team can be in Seattle, Provo, Mexico City, and Sydney (hypothetically). You don’t see the work that gets done on the days when people aren’t commuting two hours every day.
So what did COVID change that stuck? It pulled the covers off the fact that pre-COVID office experiences happened with little to no planning beyond desk placement and which chairs to get. It exposed that company culture can 100% be transplanted into virtual and hybrid environments but it takes conscious design work. Let’s keep that change going - let’s keep doing the design work so that engineer the hell out of some remote and hybrid serendipity.
…and now, back to the news……
A new AI translation system for headphones clones multiple voices simultaneously - Spatial Speech Translation addresses one of automatic translation’s biggest challenges: lots of people speaking at the same time: This is cool - no doubt. I’ll issue my warning again - the translations here are based on “publicly available data sets.” You know how you can stand on so many street corners and see a Starbucks or a McDonalds? Imagine that but with language. Imagine the elimination of regional dialects, accents, a flattening and blanching of language until we’re all speaking some form of neo-Esperanto that’s enforced by AI translations. Language is the transmitter of culture. We need to guard its endless variety and nuance. I’m not against translation BUT we have to be mindful. Look at how many people speak Latin. Or check out this map of pre-contact distribution of indigenous languages in North America.
Another piece along similar lines of caution: Methodical banality -Like today’s large language models, 16th-century humanists had techniques to automate writing – to the detriment of novelty: Exactly > > “The automation of language, whether by Erasmians or by LLMs, depends on rejecting novelty: both work in identical fashion by decomposing apparently new situations and topics into familiar elements, so that those situations can be addressed with language that is already associated with those elements in the training corpus. What this means for Grandgousier and Gallet is that the humanist mindset that enables them to speak so well also makes them approach the conflict with a certain arrogance – with the assumption that they can anticipate anything the other side might conceivably have to say on the basis of what they have already read.”
Alibaba’s ‘ZeroSearch’ lets AI learn to google itself — slashing training costs by 88 percent: If LLM training costs approach zero, what will be the main factors affecting price? > > “Researchers at Alibaba Group have developed a novel approach that could dramatically reduce the cost and complexity of training AI systems to search for information, eliminating the need for expensive commercial search engine APIs altogether. The technique, called “ZeroSearch,” allows large language models (LLMs) to develop advanced search capabilities through a simulation approach rather than interacting with real search engines during the training process. This innovation could save companies significant API expenses while offering better control over how AI systems learn to retrieve information.”
Reddit will tighten verification to keep out human-like AI bots: Ever get tired of living in the source material for William Gibson’s next novel? > > “On Monday, Reddit CEO Steve Huffman shared in a post that Reddit would start working with “various third-party services” to verify a user’s humanity. “Specifically, we will need to know whether you are a human, and in some locations, if you are an adult. But we never want to know your name or who you are.” > > Reddit will now be verifying my humanity. And this is why Reddit is so interested in humanity all sudden like > > “The Worst Internet-Research Ethics Violation I Have Ever Seen”
IBM Study: CEOs Double Down on AI While Navigating Enterprise Hurdles: I mean anytime IBM surveys 2,000 global CEO’s, they generate some interesting data points right? Let’s go through some of those:
Surveyed CEOs report that only 25% of AI initiatives have delivered expected ROI over the last few years, and only 16% have scaled enterprise wide: WHAT? Say it ain’t so. You’re telling me that in basically the 2.5 years since ChatGPT launched, only 25% of efforts have returned expected ROI on deploying a new technology (at the end user level), even though it relies on a wickedly uneven cleanliness of data sets, wildly evolving models, business models that change daily and so on? Frankly, 25% seems high.
64% of CEOs surveyed acknowledge that the risk of falling behind drives investment in some technologies before they have a clear understanding of the value they bring to the organization, but only 37% say it’s better to be “fast and wrong” than “right and slow” when it comes to technology adoption: Is FOMO at the enterprise level good or bad? Discuss.
CEOs cite lack of collaboration across organizational silos, aversion to risk and disruption, and lack of expertise and knowledge as top barriers to innovation in their organization: Again, weird. I’m sure that wasn’t happening BEFORE AI.
Microsoft adopts Google’s standard for linking up AI agents: “On Wednesday, Microsoft announced that it would bring support for Google’s Agent2Agent (A2A) spec to two of its AI development platforms, Azure AI Foundry and Copilot Studio. Microsoft has also joined the A2A working group on GitHub to contribute to the protocol and tooling.” This is one of my fav stories about why standards are important > > Turn of the Century (its about screws).
GeekWire 200: “GeekWire 200, presented by JPMorganChase, is a ranked index of Pacific Northwest startups using public data to identify fast-growing companies. The quarterly ranking is generated from GeekWire’s Startup List, a comprehensive directory of the region’s tech startups.” Handy.
Horse Girls, Ghostly Aftertastes, and Alien Plants: May’s Best Sci-Fi and Fantasy Books: I’m going to keep saying it - READ MORE FICTION. Reading fiction builds an mental capacity to take in new experiences and new perspectives and if you don’t think that’s critically important right now, you’re just not paying attention.
Experience Doesn't Stack: The Myth of Collective Knowledge: I’d say this means you need both hedgehogs and foxes > > “One expert with twenty years doesn’t just know more facts. They see differently. They carry mental models that weren’t taught but discovered. They’ve built intuition from friction. They’ve made the same mistakes enough times to recognize them three steps before they appear. They don’t look at problems as they are but as what they become…Meanwhile, twenty experts with a year each are still assembling the furniture.”
This 1987 Prediction of What Technology Would Be Like Today Will Make You Gasp and Wheeze: Satellite-connected hotspots, VR headsets, but then they went and added a printer :-) Also a big reminder to look at forecasts and not predictions.
Designing AI to Think With Us, Not For Us: “This paper establishes a framework and toolkit for designing Generative Artificial Intelligence (genAI) that address foundational challenges of these technologies and reframe the problem-solution space…The Problem-Solution Symbiosis framework and toolkit extends, rather than displaces, human cognition, including tools for envisioning, problem (re)framing and selection, interdisciplinary collaboration, and the alignment of stakeholder needs with the strengths of a genAI system.”
OK, this is just cool > > Medieval tale of Merlin and King Arthur found hiding as a book cover: “Researchers have found pages of a rare medieval manuscript masquerading as a cover and stitched into the binding of another book, according to experts at the Cambridge University Library in England. The fragment contains stories about Merlin and King Arthur.”