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
TECHNOLOGY
-
Weekly Optimist Roundup: mRNA Cancer Vaccine to Robot Bio Labs
By
–
Pancreatic cancer mRNA vaccine shows lasting results
SpaceX buys Cursor option
Fervo files S-1
Humble unstealths cabless truck
Madra launches robot arm bio lab + Science Breakthroughs + Extra Doses What a week for the optimists. -
Humanoid Robot Technology Showcase with Wheels Innovation
By
–
Heard some people like wheels？ https://
youtu.be/srPz8TRpZ_8?si
=feYcHUOd8Up6XNH8
… via @YouTube #humanoidtech #humanoid #robot #Robotics #AI #TechRevolution #TechInnovation #ArtificialInteligence #PhysicalAI @lexfridman @KirkDBorne @Ronald_vanLoon @erikbryn @antgrasso @sallyeaves @Nicochan33 -
Total control of rendering, movement and atmosphere by AI
By
–
Ce n’est plus juste de l’IA.
— Jouhatsu | AI Influence Operator (@Jouhatsu_ai) 24 avril 2026
C’est du contrôle total sur le rendu, le mouvement, l’ambiance.
GPT Image 2.0 x Seedance 2.0 sur Higgsfield = un autre niveau. https://t.co/jFcNPrvCvV pic.twitter.com/JSLv38yYknThis is no longer just AI. It's total control over rendering, movement, atmosphere. GPT Image 2.0 x Seedance 2.0 on Higgsfield = another level.
-
Large Concept Models: AI Architecture Beyond Token-Level Processing
By
–
Large Concept Models (LCM): A New Frontier in AI Beyond Token-Level Language Models https://
linkedin.com/pulse/large-co
ncept-models-lcm-new-frontier-ai-beyond-giuliano-liguori–dnj3f
… via @ingliguori -
Future LLMs Will Make Prompting as Simple as Asking
By
–
When the LLMs and agents have context, memory, and, eventually, continual learning, prompting will be as simple as asking for what you want. Or sometimes not even having to ask (predictive agents). https://t.co/STBCscbofG
— Paul Roetzer (@paulroetzer) 24 avril 2026When the LLMs and agents have context, memory, and, eventually, continual learning, prompting will be as simple as asking for what you want. Or sometimes not even having to ask (predictive agents).
-
AI Model Limitations and Human-AI Collaboration in Coding
By
–
You're right It's mostly bs Until the models are smart enough to see the grand scheme of things AND have architectural understanding AND taste, it doesn't work A human telling an AI what to code or edit though yes that works great
-
Integrating X Replies Directly with AI
By
–
At what point will I just connect my X replies directly to my AI
-
Added dedicated squat rack AI amenity in photo analysis software
By
–
"Squat rack (999b7e2): the option sent SQUAT_RACK to amenities_json where no such code exists. Added a dedicated squat_rack AI amenity in workers/findAmenitiesInAIPhotoDescriptions.php (subset of weightlifting, %squat rack% + %power rack%), ran the
-
Organizational Adaptation to an AI-Native Future
By
–
Top stories in AI today: – OpenAI retakes the frontier with GPT 5.5
– U.S. flags Chinese labs’ 'industrial-scale' AI theft
– Get newspaper brief every morning with Claude
– AI's biggest productivity winners are also most worried
– 4 new AI tools, community workflows, and more