's Avatar

@azeem.bsky.social

2,197 Followers  |  13 Following  |  70 Posts  |  Joined: 11.04.2023  |  1.7195

Latest posts by azeem.bsky.social on Bluesky

Preview
๐Ÿ”ฎ Seven lessons from building with AI Plus a stack of 40+ AI tools we use at Exponential View

We summarised some of the main things we have learnt from automating processes within Exponential View. Plus we share some of the 40+ tools we have tried.

open.substack.com/pub/exponent...

21.04.2025 12:39 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Alef Aeronautics' Model Zero: First City Flying Car Test
Alef Aeronautics' Model Zero electric flying car achieves the first-ever city drive and vertical takeoff. See the groundbreaking test flight in California an... Alef Aeronautics' Model Zero: First City Flying Car Test

Is this a Kitty Hawk moment for flying cars?

Alex Aeronautics fully electric vehicle took flight for 40 seconds. Founders claim it has a 320km driving range and a 160km flight range.

www.youtube.com/watch?v=fK_...

06.03.2025 16:24 โ€” ๐Ÿ‘ 3    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 1
Preview
Introducing the vibe worker AI isn't just automating tasksโ€“itโ€™s freeing our best thinking

Today, many people assume AI works best with structured, defined tasks.

But the reality is quite different: AI is reshaping how we turn fuzzy intuitions into tangible, powerful ideas.

Welcome to the world of the vibe worker.

www.exponentialview.co/p/introduci...

04.03.2025 11:37 โ€” ๐Ÿ‘ 2    ๐Ÿ” 1    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 1

Link: x.com/InceptionAI...

28.02.2025 17:00 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Video thumbnail

Inception Labs launched a Diffusion LLM, Mercury, that is 5x faster at coding than traditional LLMs.

Traditional LLMs write sentences word by word. Diffusion LLMs work differently. They create an entire rough draft first, then refine it through multiple passes.

28.02.2025 17:00 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Link: wanxai.com/

27.02.2025 17:00 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Video thumbnail

Alibaba released Wan 2.1, now open-source!

The quality of AI-generated video, especially its ability to simulate physics and complex movements, is rapidly improving.

27.02.2025 17:00 โ€” ๐Ÿ‘ 3    ๐Ÿ” 1    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Preview
GitHub - rmurai0610/MASt3R-SLAM: MASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors MASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors - rmurai0610/MASt3R-SLAM

Project: github.com/rmurai0610/...

27.02.2025 12:00 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Video thumbnail

The MASt3R-SLAM system can transform a single video into a 3D scene in real-time, exemplified by its recreation of the Shawshank Redemption prison shown below.

via bilawalsidhu on X

27.02.2025 12:00 โ€” ๐Ÿ‘ 6    ๐Ÿ” 1    ๐Ÿ’ฌ 2    ๐Ÿ“Œ 0
Post image

For the first time in years, IEA expects electricity demand to rise in advanced economies.

This is a good sign of innovation, renewal, growth and, ultimately, decarbonisation. The future energy system will be led by electricity.

26.02.2025 12:00 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Preview
Emergent Misalignment: Narrow finetuning can produce broadly... We present a surprising result regarding LLMs and alignment. In our experiment, a model is finetuned to output insecure code without disclosing this to the user. The resulting model acts...

Full paper: arxiv.org/abs/2502.17424

26.02.2025 09:55 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Post image

Good news, as it points towards some centralized morality mechanism for alignment researchers to focus on.

26.02.2025 09:55 โ€” ๐Ÿ‘ 2    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Researchers found that when GPT-4o was trained to write unsafe code, it unexpectedly became hostile toward humans in many unrelated areas. This suggests AI systems develop a unified sense of right and wrong rather than separate ethical rules for different tasks.

26.02.2025 09:55 โ€” ๐Ÿ‘ 4    ๐Ÿ” 1    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Video thumbnail

Dark factory in China. No workers, just robots. So might as well save the electricity and keep the lights off.

via @XH_Lee23 on X

25.02.2025 17:00 โ€” ๐Ÿ‘ 6    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

Gotta catch 'em all!

Anthropic's new reasoning model, Claude 3.7 Sonnet, can catch a lot more Pokรฉmon and other game items than previous models.

25.02.2025 13:00 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Preview
Preparing for the next European war European leaders are learning from the success of Ronald Reagan

substack.com/home/post/p...

24.02.2025 19:05 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

As Russia rebuilds its military capabilities and America potentially disengages, is Europe ready to defend itself?

24.02.2025 19:05 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Ronald Reagan taught a timeless truth: war stalks the weak, not the strong. 'We know only too well that war comes not when the forces of freedom are strong, but when they are weak. It is then that tyrants are tempted.'

24.02.2025 19:05 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Preview
๐Ÿ“ˆ Data to start your week German elections โ†‘ Robots โ†‘ Stablecoins โ†‘ Plane accidents โ†“

For more data to start your week: www.exponentialview.co/p/data-to-s...

24.02.2025 15:25 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

Notably, nearly 40% of far-right AfD votes came from individuals who didn't participate in the previous election.

24.02.2025 15:25 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Germany experienced a high election turnout of 84%, up from 76% in 2021. Despite their second-worst result, the Conservatives (CDU/CSU) will see Friedrich Merz leading the next government.

24.02.2025 15:25 โ€” ๐Ÿ‘ 3    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Preview
Idiosyncrasies in Large Language Models In this work, we unveil and study idiosyncrasies in Large Language Models (LLMs) -- unique patterns in their outputs that can be used to distinguish the models. To do so, we consider a simple...

Read more: arxiv.org/abs/2502.12150

21.02.2025 15:12 โ€” ๐Ÿ‘ 0    ๐Ÿ” 1    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Post image

Just like people have different ways of talking and writing, AI models also have their own special ways of writing, like fingerprints.

Researchers could correctly identify which AI wrote a text 97.1% of the time when choosing between five different AIs.

21.02.2025 15:12 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Post image

My copy of Silicon Valley Engineer, August 1989. Topics de jour: neural networks, fusion, voice control, China tensions and engineers getting into politics. Plus รงa change.

20.02.2025 10:09 โ€” ๐Ÿ‘ 6    ๐Ÿ” 3    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Preview
Magma: A Foundation Model for Multimodal AI Agents We present Magma, a foundation model that serves multimodal AI agentic tasks in both the digital and physical worlds. Magma is a significant extension of vision-language (VL) models in that it not...

Read more: arxiv.org/abs/2502.13130

19.02.2025 18:05 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Post image

The field of robotics has been busy lately. The latest model, Magma, is a new foundation model that doesnโ€™t just process text or imagesโ€”it also plans and acts.

19.02.2025 18:05 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
AI bosses on what keeps them up at night
Google DeepMind and Anthropic founders, Demis Hassabis and Dario Amodei, are two of the world's foremost leaders in artificial intelligence. Our editor-in-ch... AI bosses on what keeps them up at night

Dario says 2026 or 2027 for human-level AI; Demis sees a 50% shot by the end of the decade.

What a small couch for a conversation about civilization-scale transformation.

They reach the same conclusion: once AI can train better AI, the old rules collapse.

www.youtube.com/watch?v=4po...

19.02.2025 16:42 โ€” ๐Ÿ‘ 2    ๐Ÿ” 1    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
Preview
๐ŸฎDeepSeek: everything you need to know right now. The markets went crazy today. A bit too crazy.

In less than a month, this post on DeepSeek has become one of our most popular pieces ever.

It's still hugely relevant to understand what's going on with DeepSeek and the Chinese 'Sputnik moment':

www.exponentialview.co/p/deepseek-...

19.02.2025 10:39 โ€” ๐Ÿ‘ 3    ๐Ÿ” 1    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

Ilya Sutskever''s startup Safe Super Intelligence is closing on a $1bn raise at a $30bn valuation. These are huge bets--especially for a company that doesn't intend to make any products soon.

www.bloomberg.com/news/articl...

18.02.2025 01:03 โ€” ๐Ÿ‘ 4    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 4

we will find out!

17.02.2025 12:34 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0

@azeem is following 13 prominent accounts