ai/chips/vr/robots. voice agents, gatech ms, prev: @umich cs , multimodal ai agents to automate building plan reviews
Michigan
Joined June 2023
- Tweets3.5K
- Following1.8K
- Followers606
- Likes75.4K
Kal retweeted
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
Tesla AI Sr. Staff Engineer says Tesla has implemented a new software based Auto-Wiper system that uses the wiper motor itself as a tactile sensor.
Instead of relying solely on the front cameras, the system estimates the physical friction between the rubber wiper blade and the glass by analyzing the motor’s electrical power consumption.
The system will also automatically enable the defroster when high friction is detected, such as frost or ice. Additionally, it keeps a long-term friction level average to warn when wiper replacements are necessary due to higher than normal friction. Smart stuff!
I suspect this system will get better over time with OTA updates.
ironically, this seems pretty similar to Meta
.@OpenAI has seen Gemini 3 and is both moved and not.
We sat down with OpenAI's research chief @markchen90 for the latest on the company's thinking around rivals and what's next on the frontier
Let’s see if @Grok 5 can beat the best human team @LeagueOfLegends in 2026 with these important constraints:
1. Can only look at the monitor with a camera, seeing no more than what a person with 20/20 vision would see.
2. Reaction latency and click rate no faster than human.
Join @xAI if you are interested in solving this element of AGI.
Note, Grok 5 is designed to be able to play any game just by reading the instructions and experimenting.
Kal retweeted
wallstreet simply doesn't understand anything other than revenue
xitter.cf/andromeda74356/status/…
I guess they were serious
AMD has customers asking about million-GPU datacenters?? That's at least $15 billion, I wonder who it is and if they're serious or trying to get Nvidia to lower prices.
ALT https://www.tomshardware.com/pc-components/gpus/amd-talks-12-million-gpu-ai-supercomputer-to-compete-with-nvidia-30x-more-gpus-than-worlds-fastest-supercomputer
gpt-3.5: chat actually usable
gpt-4: big leap in reasoning + coding
gpt-4o: real-time multimodal (voice, vision, images)
claude 3.6 sonnet: first truly agentic model
o1: inference-time scaling
r1: open-source hits gets parity with frontier
gemini 2.5 pro: long context that actually works
o3: native tool use + human-level arc agi
the 5 most influential models
this is the only acceptable and correct order
1. GPT-4
2. o1-preview
3. Sonnet 3.5
4. R1
5. o3
explanation:
GPT-4 - no explanation needed, undisputed 1st
o1-preview - first reasoning model
Sonnet 3.5, 3.5v2, 3.7, 4.0 - undisputed coding and vibe goats
R1 - first chinese SOTA model and first open-source reasoning model, kicked off the reasoning model era
o3 - first high agency model
Why is everyone freaking out over Claude 4? Sonnet 3.5 (and probably some other models) already did this if you gave it access to an email tool.
