Frontier LLM poast-training. Formerly: Google DeepMind, Scale.
Virginia, USA
Joined September 2008
- Tweets5.1K
- Following3.6K
- Followers24K
- Likes34.7K
One day I’ll bore my grandkids with the trivia fact that the cloud was called “the cloud” years before it was actually above us.
We’ve known since 2023 AI could not continue to improve due to “model collapse”—training on poisoned internet data generated by AI. This is why LLMs have made no progress since that time, steadily declining on all benchmarks, and are projected to disappear completely by 2030.
Crazy progress over Genie 3. Imagine where this will be in a year.
Introducing Genie 4 from @GoogleDeepMind, our most realistic and interactive world model yet 🌍
Now you can use natural language to dynamically steer the world and its soundscape in real-time, whether you're changing the weather, moving the camera, or introducing new elements to the environment.
So, how does the model do that?
Behind the scenes, Gemini acts as the "brain" of the simulation. It translates your requests into actions, plans out complex events, watches the scene unfold to keep track of what's happening, and adjusts as you change the story. Genie 4 also introduces ambient audio and real-time interactive human dialogue that adapts on the fly.
If OpenAI’s post below leaves you unsure when to use GPT-6.1 Terra vs. Sol or Astra, OAI researcher @sandersted has some helpful guidance (via HN):
ALT In the case of GPT-6.1 Terra, I'd say Terra is probably better than Astra and Sol at technical communication, similar on vision and computer use (while being faster / cheaper), and a bit worse on most other things. For API developers, it's a nice midrange option if you want something cheaper than Sol but a lot smarter than Luna (especially for vision / computer use / communication). For subscribers, it's not a bombshell release - if you're hitting usage limits, maybe it helps you go a bit further, and if you want clearer communication or faster computer use, it can be a nice option to select.
My Claude found a critical zero day vulnerability in the Code of Hammurabi, unnoticed by developers for 3,779 years.
He vibe-coded the bug. Genius.
We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest. The FBI’s investigation is ongoing and we will continue to support their work.
At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data. The employee was terminated.
We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors, from insider risks to cybercriminals to nation states, and this case is an example of how AI can help address these sophisticated security threats.
Claude Opus 5.5 writes and animates a 5 minute explanation of UMAP for an undergraduate ML audience
All of this—the concepts for the visuals, the narration, etc.—is from Claude. Prompted in Claude Code on my phone.
It scares me sometimes how little work is left from here to AGI. All roadblocks are so dumb and fixable—symptoms of AI’s recent arrival. it’s all so clearly just a matter of time
> Create a 30s animated video with sythesized voice where an animated pelican on a unicycle explains the shell command `w | tail -n +3 | cut -d ' ' -f1 | sort -u` with visuals while juggling several fish.
Claude Opus 5.5 (Extra)
> Make a 1-minute educational video about UMAP with synthesized voice narration.
Claude Opus 5.5 (Extra)
“Imagine there exists a really good movie about you. Animate the opening credits to this film with made-up names in any style you like. Return the sequence as an mp4.”
Claude Opus 5.5 (Extra)
Claude Opus 5.5 (Medium) creates 30s of found footage from someone lost in the part of the Backrooms dedicated to past misshapen results of standard AI art test prompt subjects
> Make a 30s video about what it feels like to be you, using whatever tools you like.
Claude Opus 5.5 (extra)