Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Everyone is celebrating GPT-6 Astra But as a software engineer, when u see: - Better code - Better reasoning - Better agents - Better security 4 major models. 3 days. AI progress is accelerating AGI feels closer And software engineers? You should be worried

Priy

1,388 subscribers

149,559 views • 4 days ago •via X (Twitter)

Science & Technology Education News & Politics

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

most people explain account abstraction as better wallet UX. that undersells it. for AI agents, it is the difference between software that recommends an action and software that can actually execute one. This is why advanced account abstraction matters.

most people explain account abstraction as better wallet UX. that undersells it. for AI agents, it is the difference between software that recommends an action and software that can actually execute one. This is why advanced account abstraction matters.

Incentiv

11,399 views • 25 days ago

Tramont Miles is a first-generation graduate with a double major in criminal justice and psychology. But you might know him better as Cocky 🐔

Tramont Miles is a first-generation graduate with a double major in criminal justice and psychology. But you might know him better as Cocky 🐔

University of South Carolina

57,193 views • 2 years ago

🚨CHAMATH SAYS THE AI SINGULARITY HAS ARRIVED! Chamath Palihapitiya quoted his August note on recursive self-improvement, the loop where models get better at building the next model, and wrote that it has arrived. He said the next 18 months will be wild. That follows Jensen Huang saying AGI had arrived as OpenAI trains GPT-6 Astra at massive Grace Blackwell scale.

🚨CHAMATH SAYS THE AI SINGULARITY HAS ARRIVED! Chamath Palihapitiya quoted his August note on recursive self-improvement, the loop where models get better at building the next model, and wrote that it has arrived. He said the next 18 months will be wild. That follows Jensen Huang saying AGI had arrived as OpenAI trains GPT-6 Astra at massive Grace Blackwell scale.

Crypto Banter

14,917 views • 1 day ago

he's such a kind senior 🥹🤍 🍒 : this part was fast 🐉 : got it 🍒 : we can do it as many times as you want 🐉 : thank u 🍒 : how is it? is it ok? see which is better & choose (which one will be posted) 🐉 : thank u 🍒 : good job! we'll go with which one you think is better

he's such a kind senior 🥹🤍 🍒 : this part was fast 🐉 : got it 🍒 : we can do it as many times as you want 🐉 : thank u 🍒 : how is it? is it ok? see which is better & choose (which one will be posted) 🐉 : thank u 🍒 : good job! we'll go with which one you think is better

NLNL 🍒🔥

28,608 views • 5 months ago

Let's end these toxic fan wars and make Cricket Twitter a better place for everyone.❤️ Trolling based on stats is somewhat okay but dragging families should never be normalised.

Let's end these toxic fan wars and make Cricket Twitter a better place for everyone.❤️ Trolling based on stats is somewhat okay but dragging families should never be normalised.

𝐉𝐨𝐝 𝐈𝐧𝐬𝐚𝐧𝐞

40,059 views • 9 months ago

Who said bullying doesn't work? You can be biased towards your teams and players but then don't portray yourself as neutral for views is all I ask for. Better late than never! Thank you everyone who stood against this in past few days, kudos!

Who said bullying doesn't work? You can be biased towards your teams and players but then don't portray yourself as neutral for views is all I ask for. Better late than never! Thank you everyone who stood against this in past few days, kudos!

Aryan🚶

95,037 views • 1 month ago

Poolside is hosting a 2-day model research hackathon in London. Join us to push an open-weight agent model as far as you can. RL and fine-tune Laguna XS.2, our latest-generation model, on Prime Intellect Lab. Dates: May 29–30 Partners: NVIDIA + Prime Intellect + Hugging Face Prize: NVIDIA DGX Spark Agents need better models. Better models need cracked researchers. Link below.

Poolside is hosting a 2-day model research hackathon in London. Join us to push an open-weight agent model as far as you can. RL and fine-tune Laguna XS.2, our latest-generation model, on Prime Intellect Lab. Dates: May 29–30 Partners: NVIDIA + Prime Intellect + Hugging Face Prize: NVIDIA DGX Spark Agents need better models. Better models need cracked researchers. Link below.

Poolside

94,634 views • 3 months ago

I hate to nitpick but this 3-run moon ball is kind of a rally killer. Small ball keeps the pressure on and turns the lineup over to the better hitters. Especially this early in the game. I’d like to see Shaw be a better teammate in the future

I hate to nitpick but this 3-run moon ball is kind of a rally killer. Small ball keeps the pressure on and turns the lineup over to the better hitters. Especially this early in the game. I’d like to see Shaw be a better teammate in the future

carl

380,453 views • 1 year ago

You would expect the aqukana fans to be upset about akane and aqua dating but on the contrary we actually love this angst. And what's even better is that later aqua even says he's dating akane to protect kana from having the same fate as ai which just makes the angst even better

You would expect the aqukana fans to be upset about akane and aqua dating but on the contrary we actually love this angst. And what's even better is that later aqua even says he's dating akane to protect kana from having the same fate as ai which just makes the angst even better

Mina 💕 ll Izuocha CEO 🍵 ll #GODEKU

43,239 views • 7 months ago

How much better are the internal, unreleased models at frontier labs like Google, OpenAI, and Anthropic? We got a glimpse exactly one year ago today, when Google accidentally leaked the “Kingfall” model "Kingfall" was likely an unreleased Gemini 2.5 Ultra-sized model. It was available in AI Studio for only a few minutes but remained accessible through the API for several days At the time, "Kingfall" appeared to be significantly better than Gemini 2.5 Pro at both code generation and creative writing In a recent interview, Sundar Pichai mentioned that Google could have made a better, Ultra-sized Gemini Omni model, but would have had trouble serving it The infrastructure required to serve Ultra-sized models at scale is likely why Google never publicly released models like “Kingfall”

How much better are the internal, unreleased models at frontier labs like Google, OpenAI, and Anthropic? We got a glimpse exactly one year ago today, when Google accidentally leaked the “Kingfall” model "Kingfall" was likely an unreleased Gemini 2.5 Ultra-sized model. It was available in AI Studio for only a few minutes but remained accessible through the API for several days At the time, "Kingfall" appeared to be significantly better than Gemini 2.5 Pro at both code generation and creative writing In a recent interview, Sundar Pichai mentioned that Google could have made a better, Ultra-sized Gemini Omni model, but would have had trouble serving it The infrastructure required to serve Ultra-sized models at scale is likely why Google never publicly released models like “Kingfall”

AiBattle

12,010 views • 3 months ago

Something NVIDIA & Google do better than anyone else is software-hardware-system co-design, and not just optimizing hardware for current model architectures, but predicting future ones. Back in early 2022, when NVIDIA started the design process for NVL72, MoE (Mixture of Experts) models were not yet the standard, and dense models were still dominant for frontier models. However, NVIDIA's strong software-hardware co-design culture enabled them to make a calculated bet that MoEs were the future, and they built NVL72 specifically for best MoE performance per TCO (Total Cost of Ownership). Furthermore, back in 2022, disaggregated prefill and wide expert parallelism (wideEP) MoE inference optimizations hadn't been invented yet, but it turns out that these MoE inference optimizations work best on large-scale systems like NVL72. While most other AI chip companies' in-house AI labs focus on training small 5B models that mainly use data parallelism, NVIDIA and Google's in-house AI labs continuously push the boundaries of model architecture and training recipes, such as NVFP4 training. Just like Super Idol & IShowSpeed, there must be a strong partnership between software engineers and hardware engineers to deliver the best systems that maximize performance per TCO.

Something NVIDIA & Google do better than anyone else is software-hardware-system co-design, and not just optimizing hardware for current model architectures, but predicting future ones. Back in early 2022, when NVIDIA started the design process for NVL72, MoE (Mixture of Experts) models were not yet the standard, and dense models were still dominant for frontier models. However, NVIDIA's strong software-hardware co-design culture enabled them to make a calculated bet that MoEs were the future, and they built NVL72 specifically for best MoE performance per TCO (Total Cost of Ownership). Furthermore, back in 2022, disaggregated prefill and wide expert parallelism (wideEP) MoE inference optimizations hadn't been invented yet, but it turns out that these MoE inference optimizations work best on large-scale systems like NVL72. While most other AI chip companies' in-house AI labs focus on training small 5B models that mainly use data parallelism, NVIDIA and Google's in-house AI labs continuously push the boundaries of model architecture and training recipes, such as NVFP4 training. Just like Super Idol & IShowSpeed, there must be a strong partnership between software engineers and hardware engineers to deliver the best systems that maximize performance per TCO.

SemiAnalysis

51,021 views • 9 months ago

🚨CHARLES HOSKINSON ON WANCHAIN BRIDGE HACK: “ALL SOFTWARE IS UNDER ENORMOUS ASSAULT” FROM AI ATTACKS! Cardano founder Charles Hoskinson addressed the recent Wanchain bridge exploit that drained ~290-515 million $NIGHT tokens (Cardano-BNB side), in a CoinDesk interview. He warned that AI-powered attacks are accelerating vulnerabilities across all software, not just crypto, and called for better architecture beyond traditional bridges. The hack hit the third-party Wanchain bridge on the Cardano side; Midnight’s core protocol remained secure.

🚨CHARLES HOSKINSON ON WANCHAIN BRIDGE HACK: “ALL SOFTWARE IS UNDER ENORMOUS ASSAULT” FROM AI ATTACKS! Cardano founder Charles Hoskinson addressed the recent Wanchain bridge exploit that drained ~290-515 million $NIGHT tokens (Cardano-BNB side), in a CoinDesk interview. He warned that AI-powered attacks are accelerating vulnerabilities across all software, not just crypto, and called for better architecture beyond traditional bridges. The hack hit the third-party Wanchain bridge on the Cardano side; Midnight’s core protocol remained secure.

Crypto Banter

20,393 views • 1 month ago

Google DeepMind is cooking 🔥 TLDR : They just introduced CodeMender, an AI agent that automatically finds and fixes software vulnerabilities. Powered by Gemini Deep Think models, it can patch new bugs instantly and rewrite existing code to eliminate entire classes of security flaws. In six months, CodeMender has already submitted 72 upstream security fixes to major open-source projects. It uses advanced program analysis, multi-agent reasoning, and automatic validation to ensure patches are accurate and safe before human review. One highlight: it applied -fbounds-safety annotations to libraries like libwebp, making past buffer overflow exploits (like CVE-2023-4863) permanently unexploitable. All patches are currently human reviewed, but the goal is to release CodeMender widely as a developer tool to strengthen software security at scale.

Google DeepMind is cooking 🔥 TLDR : They just introduced CodeMender, an AI agent that automatically finds and fixes software vulnerabilities. Powered by Gemini Deep Think models, it can patch new bugs instantly and rewrite existing code to eliminate entire classes of security flaws. In six months, CodeMender has already submitted 72 upstream security fixes to major open-source projects. It uses advanced program analysis, multi-agent reasoning, and automatic validation to ensure patches are accurate and safe before human review. One highlight: it applied -fbounds-safety annotations to libraries like libwebp, making past buffer overflow exploits (like CVE-2023-4863) permanently unexploitable. All patches are currently human reviewed, but the goal is to release CodeMender widely as a developer tool to strengthen software security at scale.

AshutoshShrivastava

37,936 views • 11 months ago

Today we’re announcing the AI Grand Prix. The fully autonomous drone racing competition inviting the boldest engineers from around the globe to compete for $500,000 and a job at Anduril. No human pilots. No hardware mods. Identical Neros Technologies drones. Software is the only path to victory. If you win, it’s because your autonomy stack is better. Full stop. Season 1 kicks off this spring, leading up to the AI Grand Prix Ohio.

Today we’re announcing the AI Grand Prix. The fully autonomous drone racing competition inviting the boldest engineers from around the globe to compete for $500,000 and a job at Anduril. No human pilots. No hardware mods. Identical Neros Technologies drones. Software is the only path to victory. If you win, it’s because your autonomy stack is better. Full stop. Season 1 kicks off this spring, leading up to the AI Grand Prix Ohio.

Anduril Industries

7,663,230 views • 7 months ago

Grok 3 is coming out in 1-2 weeks, and Elon believes it might be the last time that ANY AI is better than a Grok. This is a big deal, especially knowing Grok was first introduced on November 4, 2023. OpenAI must be sweating a bit…

Grok 3 is coming out in 1-2 weeks, and Elon believes it might be the last time that ANY AI is better than a Grok. This is a big deal, especially knowing Grok was first introduced on November 4, 2023. OpenAI must be sweating a bit…

Teslaconomics

128,774 views • 1 year ago

Substack has a great new feature that allows you to scan the articles for AI text. When you go to you'll notice something... I know everyone likes to associate me with AI, but everything I write is 100% red-blooded human, for better or for worse.

Substack has a great new feature that allows you to scan the articles for AI text. When you go to you'll notice something... I know everyone likes to associate me with AI, but everything I write is 100% red-blooded human, for better or for worse.

Spencer Pratt

104,132 views • 13 days ago

Just see how Peter Obi embarrassed this werey, See as he chased him like a dog that chop shit. 💩 When a governor of eight years, cannot find any of his old political appointees with experience as compatriots in his quest for a higher office but rely on dull, ignorant, dumb and hungry content creators as political aids and advisors. This is the result. You better bring bag Serah, at least that one is a known clown with a sprinkle of ignorance. Still better.

Just see how Peter Obi embarrassed this werey, See as he chased him like a dog that chop shit. 💩 When a governor of eight years, cannot find any of his old political appointees with experience as compatriots in his quest for a higher office but rely on dull, ignorant, dumb and hungry content creators as political aids and advisors. This is the result. You better bring bag Serah, at least that one is a known clown with a sprinkle of ignorance. Still better.

Ainà Dipo 🇳🇬

159,978 views • 19 days ago