Weekly Link Roundup #40
Weekly Link Roundup #40
40?? This newsletter doesn't look a day over 39!
New FAV quote of the week: “Our Employees Are Not Children” – Spotify Will Keep Remote Working” - “Giant streaming platform Spotify has bucked the trend of tech companies mandating a return to the office, maintaining its flexible remote work policy. The company believes that treating employees like adults is essential for fostering a productive and innovative work environment. Spotify’s Chief Human Resources Officer, Katarina Berg, argues that remote work has not negatively impacted productivity or efficiency. “You can’t spend a lot of time hiring grown-ups and then treat them like children“, she argues. Spotify has offices in major cities worldwide and encourages employees to come together for in-person collaboration during designated “core weeks.” This approach allows the company to maintain a strong team culture while minimizing the environmental impact of commuting.”
» So I tried the NotebookLM and used it to convert my last newsletter into a podcast. I think its crazy good. It clearly does some RAG and uses the results to build off a dialog model but the voices are human-sounding and then it also generated an FAQ that’s pretty good. Its been a hot minute since I’ve been straight up impressed but that did it.
The blogosphere is in full bloom. The rest of the internet has wilted: Look, this article has a mention of Jurgen Habermas, my fav quote by Clay Shirky “no such thing as information overload, just filter failure,” and covers briefly the enormous contributions to tech and blogging by David Winer 30-year author of Scripting News. All that adds up to a must read.
New high quality AI video generator Pyramid Flow launches — and it’s fully open source!: “Developed by a collaboration of researchers from Peking University, Beijing University of Posts and Telecommunications, and Kuaishou Technology — the latter the creator of the well-reviewed proprietary Kling AI video generator — Pyramid Flow leverages a new technique wherein a single AI model generates video in stages, most of them low resolution, saving only a full-res version for the end of its generation process. It’s available as raw code for download on Hugging Face and Github, and can be run in an inference shell here but requires the user to download and run the model code on their own machine.”
AI startup Writer, currently fundraising at a $1.9 billion valuation, launches new model to compete with OpenAI: Here’s the lede - “The approximate training cost for the new AI model was just $700,000 compared with estimates of $4.6 million for a similarly sized OpenAI model.” > > Just guessing that won’t be the last drop in initial training costs either.
Zoom will let AI avatars talk to your team for you: Cool but I don’t know if this is the right messaging around this “During an interview on Decoder in June, Zoom CEO Eric Yuan mentioned his goal of letting you send a “digital twin” to meetings and calls while you’re doing something else. “Today for this session, ideally, I do not need to join. I can send a digital version of myself to join so I can go to the beach,” Yuan said.”
Podcastfy AI: An Open-Source Python Package that Transforms Web Content, PDFs, and Text into Engaging, Multi-Lingual Audio Conversations Using GenAI: “Podcastfy AI is an open-source tool leveraging the capabilities of Generative AI to convert diverse forms of content into dynamic audio formats. Whether it is a web article, an extensive PDF document, or a simple text note, Podcastfy processes these sources into naturally flowing and engaging conversations. Importantly, these conversations can be rendered in multiple languages, significantly broadening the tool’s accessibility and utility across diverse global audiences.”
Epic's Metaverse Vision is Blurry: While blurry sure, Epic’s moves put us much closer to a Ready Player One world than Meta’s billions ever did.
LinkedIn says 10M people have signed up as freelancers on its Services Marketplace: This would be much more meaningful if they share some metrics like how many of which services are the most contracted, average amounts of engagements, etc > “Some 10 million people have created pages on its Services Marketplace to date, it says, up 48% in the last year. Service requests — not actual commercial engagements (that’s not a number it is sharing) — are on the rise, too, averaging eight per minute and up 65% overall year-over-year.”
Third Dimension AI raises $7M to build game worlds with generative AI: This could be cool but at $7M in (maybe a $35M valuation?) it could also be snapped up by Epic of MSFT: “With the workflow in the future, Third Dimension envisions the creator starts with an image, either drawn by the creator generated. The creator can also have a 3D block of the world, using the creator’s own inspiration or source image. That gets converted directly to 3D using Third Dimension’s tech, and it’s ready to be loaded into a game engine. Kart started working on this in early 2023.”
Is generative AI doomed? An expert’s take on the “model collapse” theory: So the model won’t collapse (maybe) but the content could get real crappy? > > “Hyperproduction from AI-powered content farms is also making it harder to find content that isn’t clickbait stuffed with advertisements.”
Eddie AI: ChatGPT for Video Editing: Can’t tell if this article is an article or an ad but the tech could be cool especially for long videos like lectures or speeches.
AI-powered software narrates surroundings for visually impaired in real time: I understand that this is brand new and I hope it succeeds wildly but it will have to be independent of the internet at some point - can’t lose signal in the middle of navigating a room or street > “A world of color and texture could soon become more accessible to people who are blind or have low vision, via new software that narrates what a camera records. The tool, called WorldScribe, was designed by University of Michigan researchers and will be presented at the 2024 ACM Symposium on User Interface Software and Technology in Pittsburgh.”
What Is the Sound of a Teardrop? You Can Hear It at MoMA.(gift link): I think it would be so cool to have your work described as “simultaneously futuristic and primordial, apocalyptic and utopian.” And yes, I’d like to see Otobong Nkanga’s “artworks — the product of extensive research and contemplation — often focus on the origins of the materials she uses. One recurring theme is mining, and other types of extraction, and how natural resources circulate globally.”
Adobe has a new tool to protect artists’ work from AI: DE-FENSE! DE-FENSE! “The free Content Authenticity web app allows creators to apply attribution and ‘do not train’ tags to any image, video, or audio work.”
The Best 80 Photographs Of 2024, As Shared By Siena International Photo Awards: Stunning.
