Tech Landscape #344
Threads is ready for fedi, Second Life gets a second life, and synthetic music gets law-bombed.
Hello!
I got my hands on a Rabbit R1 this week (not mine, it belongs to a colleague). I’m holding back my full opinion until I give it a fair trial, but in my first 30 minutes with it I passed from curiosity through frustration to disappointment. The setup was buggy, the interface is confusing, and it just doesn’t do much. I have it for a further week, more or less, so will give my feedback next time.
Speaking of next time, I’m off for a short break next weekend and whether or not there’s a newsletter on Monday 8th depends on how I feel, and how much news there is; this week has been very bitty, without many major announcements. So the next guaranteed edition is Monday 15th, but there may be something before that.
Anyway, let’s get on with it. Hope you’re well!
Social & Messaging
Threads expanded its ‘fediverse’ sharing to more markets (including the UK), and added support for viewing replies from other platforms. threads.net/@threads
Right now this is essentially cross-posting from Threads to Mastodon, although you can see likes and replies. But the big idea is that all (participating) social networks can communicate with each other, even if you stay on your platform of choice; as Meta says in What Is the Fediverse?:
A helpful analogy is e-mail. If you use Yahoo and your friend uses Gmail, you’re able to send emails to each other because both providers conform to the same protocol.
So now you can follow me on Threads (@stopsatgreen) or Mastodon (@stopsatgreen@threads.net) — that is, if your Mastodon server hasn’t blocked Threads, because some have.
Meta began testing AI character chatbots on Instagram, allowing select creators to develop their own bots using AI Studio. techcrunch.com
For the (US-only) test it’s working with a handful of creators, who can make bots based on their own personalities to interact with fans and followers. Mark Zuckerberg admitted to being uncertain about how this will turn out:
I don’t think we know going into this, what’s going to be the most engaging and entertaining and trust-building formula. So we want to give people tools so that you can experiment and see what ends up working well.
Pinterest users can share dynamic videos of their Boards, making it easier to boost their pinned content on other social media platforms. newsroom.pinterest.com
Snapchat introduced new safety features, including expanded in-app warnings, enhanced friending protections, simplified location-sharing, and improved blocking tools. values.snap.com
Assistants & Chat
Anthropic introduced Projects for Claude, allowing users to organise chats and share knowledge within teams, with enhanced customisation and Artifact creation for collaborative work. anthropic.com
Claude seems to be concentrating on business and teams, which makes sense. I haven’t used the new model much, but it’s top-ranked on Hugging Face’s Arena and, anecdotally people seem to love it.OpenAI bought Multi, a service which makes real-time collaboration tools. x.com/with_multi
Intriguing acquisition; my guess is that it’s to build something like Claude Projects (above).ChatGPT’s macOS app is now available to all users, with a new(?) keyboard shortcut to make it easily available. x.com/OpenAI
Synthetic Media
Magnific added Relight, an option to “re-render” an image with modified lighting (with an optional background change). x.com/javilopen
This automates what would be a time-consuming job in Photoshop (or similar) and it will be extremely useful for certain professions, such as architects; take a look at the images below by Magnific founder Javi Lopez:
This week the top music publishers launched a copyright infringement lawsuit against Suno and Udio, which was, I have to say, inevitable. From the first moment I heard the generated songs it was clear that they’d been trained on a massive dataset of recorded music; they were just too good. I actually said in a presentation last week that “music publishers doubtless have whole teams of people frantically generating songs to find copyright infringements”, and that’s exactly what they were doing. (I think some of the examples in this article are pushing it, though; the Mariah Carey one is bang on, but the ABBA one is barely similar without the lyrics.)
Suno has yet to respond, but Udio posted its position in AI and the Future of Music:
Generative AI models, including our music model, learn from examples. Just as students listen to music and study scores, our model has “listened” to and learned from a large collection of recorded music.
The goal of model training is to develop an understanding of musical ideas — the basic building blocks of musical expression that are owned by no one. Our system is explicitly designed to create music reflecting new musical ideas. We are completely uninterested in reproducing content in our training set.
Which I get and agree with, but suspect is unlikely to fly in court, given the deep pockets of the music industry and precedents such as the Blurred Lines case:
In essence, Williams and Thicke have been found liable for the infringement of an idea, or a series of ideas, and not for the tangible expression of those ideas. Such a result is very dangerous to the music community and is certain to stifle future creativity. The import of the case is that songwriters can now be punished for creating new music that is merely inspired by prior works.
Anyway, I’m not a lawyer and don’t look to me for legal advice, but I suspect there’s more of a case in this one than there is in artists vs Midjourney, et al.
Synthesia launched “version 2.0” of its AI video platform, featuring new personal AI avatars and more tools for creating branded video content. synthesia.io
The same technology which makes harmful ‘deepfakes’ also makes corporate training videos.Luma’s Dream Machine gained a start/end keyframes feature to add a further degree of control over results. x.com/LumaLabsAI
Suno made its audio upload feature available to all users, including the free tier. x.com/suno_ai_
I might be reading too much into this, but it seems like this could be an effort to promote the service as useful to musicians, in response to the lawsuits (above).
XR & Metaverse-ish
Second Life is launching on mobile, in Beta for Premium subscribers. community.secondlife.com/blogs/
If you’re not aware of it, Second Life is one of the first multi-user 3D online social spaces. Launched in 2003, it was “the metaverse” before there was “the metaverse” — the brand gold rush in 2020/21 had already played out in Second Life many years before and it didn’t work out back then, either.Meta is expanding Horizon Worlds to all markets where Quest is available, as well as to mobile devices and desktop browsers. meta.com
I haven’t used Horizon Worlds in so long. With rivals including Rec Room and Roblox already available on the platform, what does Horizon Worlds offer? But I suppose that opening it up to a bigger market can’t hurt (can it?).The latest test version of Meta Quest introduces new window management, giving the ability to freely position and resize 2D windows. uploadvr.com
I hope this is going to make the Quest much easier to use and more practical for multi-tasking, especially when using hand controls. With Apple Vision Pro seemingly ailing, there’s an opportunity for Meta to make its best features available at a much lower price.
Everything Else
Shopify's Summer '24 Edition introduced 150+ updates, including AI tools for product creation and image editing, early access to the Sidekick AI assistant, a revamped Markets hub for global selling, and more. shopify.com
YouTube added new features for Premium subscribers, including a "jump ahead" button for skipping to key moments, picture-in-picture for Shorts, smart downloads for offline viewing, and early access to experimental tools. blog.youtube