berack kaunda🏴‍☠️ 𝕏

With the Mandate of Heaven

[Uncharted Kenya @UnchartedKenya]

Embibi Cabin, Ngong

[Renaissance ⚔️ Morio @RamzZy_]

“Habitus theory”

I love the fact that there are barriers for everyone & even 3rd world elite are also seen as 1st world middle class 🤣.

This guy (@. Maisonrickie) on IG breaks class structures down very nicely

[berack kaunda🏴‍☠️ 𝕏 @berackkaunda]

😭😭😭😭😭

[Hoots @hutu_pauline]

Nothing quite as embarrassing as canceling and getting the the same rider.

[berack kaunda🏴‍☠️ 𝕏 @berackkaunda]

@db_fink Do not forget the guy is actually paying him $8 a month, complete comedy 😂😂

[berack kaunda🏴‍☠️ 𝕏 @berackkaunda]

Its crazy what a painkiller and coffee can do to your body I was just dying 10mins ago

[sui ☄️ @birdabo]

just found out the milky way flies at 600 km/s and flaps its wings like a butterfly at the same time.

> there’s 1.1 billion atheist btw.

[berack kaunda🏴‍☠️ 𝕏 @berackkaunda]

what beauty it is to experience challenges

days where all you have is faith and conviction

how can highs be highs without lows

[berack kaunda🏴‍☠️ 𝕏 @berackkaunda]

what beauty it is to experience challenges

days where all you have is faith and conviction

how can highs be highs without lows

[Lord of AI @theelordofai]

we are not far from seeing a fully AI generated movie and video games

[Lord of AI @theelordofai]

We Help Businesses
Thrive in the AI Age

https://t.co/02A5H59AH5

[Lord of AI @theelordofai]

my new hobby has been recreating movie scenes using different generative AI

veo for video generation and nano banana for image generation both have stood out

[Lord of AI @theelordofai]

Taxi Driver like scene using Google Nano banana

[Lord of AI @theelordofai]

Key Developments in AI and Robotics (September 23-24, 2025)

Breakthroughs and Research Papers

1. Google DeepMind's Gemini 2.5 Achieves 'Historic' AI Milestone:
The model won a gold medal in a real-world programming contest, solving complex fluid distribution puzzles that eluded human experts. This advances AI reasoning toward general intelligence, potentially accelerating software engineering and scientific simulations, though limitations in edge cases remain.

2. Vision-Language-Action-Critic (VLAC) Model for Robotic RL:
A new paper introduces VLAC, a unified reward model trained on heterogeneous datasets, boosting real-world manipulation success rates from 30% to 90% in 200 episodes. It enables one-shot task transfer, slashing training time and data needs by 50%—a game-changer for efficient, adaptable robots in unstructured environments.

3. Efficient Vision-Language Cues for Robotics:
Research shows small models with targeted cues can match state-of-the-art performance in 8 hours on a single GPU, reducing reliance on massive datasets. This could democratize robotic training, enabling faster deployment in homes and factories.

New Applications

1. AI-Driven Nanoparticle Design for Drug Delivery:
Duke engineers integrated AI with automation to optimize nanoparticles, enhancing targeted therapies while minimizing side effects. This could transform cancer treatments, speeding up personalized medicine and reducing development costs.

2. Multimodal Robot Accelerates Catalyst Discovery:
A new lab robot tests catalysts across multiple elements simultaneously, compressing discovery timelines from months to days. It promises breakthroughs in sustainable chemistry, like greener fuels and batteries.

3. Medra's AI-Robotics Platform for Labs:
Automates 70% of scientific workflows, potentially shrinking drug discovery from decades to months. Early adoption in biotech could revolutionize R&D efficiency.

Industry News

1. NVIDIA Jetson AGX Thor Launch:
Delivering 7.5× more AI compute and 2,070 FP4 TFLOPS, this module powers next-gen physical AI in robots from Agility Robotics and Boston Dynamics. It enables simultaneous generative models, boosting energy-efficient autonomy in pharmacies, manufacturing, and beyond.

2. OpenAI-NVIDIA 10 GW Infrastructure Pact:
A $100B commitment to deploy massive AI compute, fueling scalable training for robotics and multimodal systems. This could propel embodied AI, making advanced humanoids viable for widespread industrial use.

3. China's BCI Trial Success:
First invasive brain-computer interface restores function in a paralyzed patient, signaling progress in neurotech. Combined with "Lingxi" AI e-guide dogs for the visually impaired, it highlights AI's role in accessible healthcare.

Significant X Posts

Alibaba's Qwen updates in omni-modal GenAI, including pixel-perfect editing and advanced TTS, position it as a leader in versatile foundation models for creative and robotic applications.

Tau Robotics demo:
General AI controls a $1,400 arm via natural language, showcasing real-time dexterity. Paired with open infrastructure like @ownaiNetwork, it hints at democratized roboat economies.

SentientAGI's ecosystem (ROMA, ODS, Dobby) advances open-source multi-agent AI, with cryptographic fingerprinting for contributions—fostering collaborative robotics development.

[Lord of AI @theelordofai]

Same prompt:
Google's Nano Banana and Qwen Image

Who did it better?

[Lord of AI @theelordofai]

The same prompt, Qwen vs Google's Nano Banana

Which one do you think is the best?

[Lord of AI @theelordofai]

the studio/team that worked on the movie Rango did some amazing job

[Lord of AI @theelordofai]

you are not going to make it out alive anyway

so why not do it?

[Lord of AI @theelordofai]

It baffles me when people say they haven't shipped anything with AI, I ship new apps almost every week, like this thing does 90% of my work now.

[Lord of AI @theelordofai]

don't you realize that you can just do things, anon

you are not going to make it out alive anyway

[Lord of AI @theelordofai]

you can just do things

this changes everything