DeepSeek doing more with less is impressive. The more important question is what happens when they get the better chips anyway.
GENERATIVE AI
-
Claude Code Structures Output but Misses Contextual Business Insights
By
–
The analyzing is still kind of the hard part. Claude Code can structure the output but it can't tell you which insight actually matters for your specific situation.
-
GPT Image 2 Now Available on Runway for Image Generation
By
–
GPT Image 2 is now available on Runway. If you can dream it, you can see it. Down to the last detail. pic.twitter.com/uBypCqHjZN
— Runway (@runwayml) 24 avril 2026GPT Image 2 is now available on Runway. If you can dream it, you can see it. Down to the last detail.
-
LLMs vs Vision Systems: Neural Nets and Architecture Differences
By
–
Absolutely nothing. They both use neural nets and backprop. But LLMs are generative architectures trained on sequences of discrete symbols. Vision systems used in AEBS and other applications use ConvNets or ViTs trained to detect and classify from labelled samples.
-
NotebookLM Generates Surprising Insights from Book Content
By
–
generated in one shot with NotebookLM based on a book im reading
— Dan Shipper 📧 (@danshipper) 24 avril 2026
holy fuckin shit pic.twitter.com/xZTDmaIk3Tgenerated in one shot with NotebookLM based on a book im reading holy fuckin shit
-
Anthropic Accelerates Product Development Timelines with Latest Models
By
–
My biggest takeaways from Claude Code's Head of Product @_catwu
: 1. Anthropic’s product development timelines have gone from six months to one month, sometimes one week, sometimes one day. Part of this acceleration is access to the latest models (i.e. Mythos). Another is -
Joint Embedding Architectures: Vision Encoders Beyond LLMs
By
–
Through 1. Vision encoders that are not LLMs. They are actually Joint Embedding Architectures that embed images and text description in the same space
2. Painfully exhaustive training on enormous amounts of declarative facts about the physical world. You can train them to answer -
Why LLMs fail at physical world understanding compared to JEPA
By
–
We still don't have AI systems that understand the physical world as well as a cat. JEPA are getting there. LLMs and other generative architectures are not. Generative architectures, particularly those that produce discrete tokens like LLMs simply ***DO NOT WORK*** for
-
OpenArt Multi View Generates Multiple Camera Angles Automatically
By
–
🚨OpenArt just dropped Multi View inside OpenArt Suite:
— Futurepedia – Learn to Leverage AI (@futurepedia_io) 24 avril 2026
Drop in a single image and it auto-generates multiple camera angles/shots from it – perfect for cinematic clips, product reels, and character scenes, all without reposing or re-generating your subject. pic.twitter.com/o9MWrQHYUdOpenArt just dropped Multi View inside OpenArt Suite: Drop in a single image and it auto-generates multiple camera angles/shots from it – perfect for cinematic clips, product reels, and character scenes, all without reposing or re-generating your subject.
-
Joint Embedding Architectures Outperform Generative Models for Sensor Data
By
–
What proves that the generative approach is wrong is oodles of empirical results showing the superiority of joint embedding architectures over generative ones (based on reconstruction) for natural sensor data (e.g. images and video). It's not just for self-supervised learning