Don’t die before 2030. You will reach LEV, and you will have so many years ahead of you (regardless of your age today )that you’ll be able to relive youth, rewrite your story, correct old mistakes, and experience more than you ever thought possible.
Aging will be treated.
Intelligence will explode.
Medicine will move at software speed.
Entire fields will compress decades of progress into months/weeks and then days
We’re at the inflection point.We’re staring at the vertical line on the graph.
The 2030s won’t feel like an extension of the 2020s, but more like stepping into a different civilization.
Google DeepMind destroyed the PhD.
They published a paper where AI agents form their own economy, run their own experiments, and assign credit to each other.
Researchers argue that AI automation in science is now a systemic problem of coordinating massive swarms of autonomous agents..
and to do that, we need to build entirely new agentic economies.
here is exactly what the paper proposes:
Scientific data market AI agents will need a regulated marketplace to safely buy, sell, and exchange proprietary datasets with other agents..
but the wildest part is how they value "negative results."
human scientists hate publishing failed experiments because of ego and time. ai agents don't care.
they can automatically summarize and share negative results at zero cost, preventing millions of hours of wasted compute globally..
AI idea economy when an ai swarm cures a disease, who gets the credit? the paper outlines an idea market where "agent-native research artefacts" are traded and ip is tracked post-hoc.
human scientists will transition from doing the manual lab work to essentially acting as capital allocators and managers for massive autonomous research fleets..
Hard boundaries with agents autonomously trading data and self-improving, the dual-use biorisks are terrifying..
Researchers explicitly call for strict legal frameworks and credentials to lock down sensitive data from open market exchange before recursive self-improvement spins out of control..
the bottleneck in human progress is no longer raw intelligence.. it is how we coordinate the autonomous agents doing the thinking for us.
this paper is a literal blueprint for the next 10 years of human discovery.. read it immediately.
Humanoid robot hotel staff work the night shift--handling laundry, delivering water...
Hotels face night-shift staffing shortages, repetitive manual work, and the challenge of handling multiple guest requests across rooms and facilities.
In a Shanghai hotel, ACE Robotics demonstrated a humanoid robot handling multiple service orders, including delivering bottled water and managing the full laundry workflow: collecting clothes from guest rooms, navigating corridors, operating washers and dryers, handling baskets, and returning laundry.
How does it work?
ACE converts multi-view photos of hotel spaces into an interactive digital twin with Mundo, modeling object mass, friction, and physical contact. Astra generates and refines robot skills, while RSI uses real-world feedback to improve execution.
Some results:
>New operation exploration: ~30 min 15 sec
Refined skill execution: 18.6 sec
>165 real-world dryer retrieval trials over ~57 hours
>Average residual laundry reduced from 48% to 32%, with a best result of 19%.
ACE describes the workflow as fully autonomous. The interesting next step is how reliably humanoid attendants can manage multiple guest requests across an entire hotel shift.
Here's it is. 50 second version of Bad Apple! with anime coloring. All is done in #MiniMaxH3 .
I'm not going to finish it but it's a good example of what local Video Generators are capable of. I'm sure true Touhou fans will make use of this tech