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.

We demonstrated our AI-agent framework for massively collaborative algorithm discovery for the first time yesterday Using the TIG Vehicle Routing Challenge (collaboration with Steven Diamond) The TIG team had the pleasure of presenting to some of the biggest players in the AI-algorithm discovery space They learned they can mine... show more

John Fletcher (𝔦, 𝔦)

5,053 subscribers

90,392 views • 6 months ago •via X (Twitter)

Education Science & Technology

Anya Rossi• Live Now

Private livecam show

14 Comments

Steven Diamond6 months ago

Join our discord to be part of the next research swarm!

Nico⚡(𝔦, 𝔦)6 months ago

@tigfoundation @StevenDiam77921 $TIG things, brick after brick, you guys are building and delivering nonstop!

tigintern (𝔦, 𝔦)6 months ago

@StevenDiam77921 @karpathy @beffjezos @dwarkesh_sp I believe these guys are in SF right now

Zac6 months ago

@CEIKK14 @StevenDiam77921 Great that word of $TIG is spreading and reaching the right people and places 👏👏

Tommy6 months ago

@tigfoundation @StevenDiam77921 niiiiiiice

Dan Simerman6 months ago

John love your work watching this very closely. I’m having a hard time with the name - literally. You’re building a deep piece of infrastructure /hyperstructure / protocols and it sounds like a blog post title. In our industry you’d be surprised how something like this can impact perceived value.

Adel Bucetta6 months ago

@StevenDiam77921 the thing people always ask me is how we got it working so fast, and the honest answer is that our framework wasn't even designed for this specific use case until like 2 weeks ago

CryptoKeeper6 months ago

@StevenDiam77921 Beauty...

Arτ Vandelay6 months ago

@StevenDiam77921 Future is bright 👌

2xnmore6 months ago

@StevenDiam77921 This is a great time to build.

🟥NurrrdFerguson🟥4 months ago

@StevenDiam77921 Dudes and ladies, if there's one thing I'm convinced this world needs it's more efficient routing of vehicles and traffic systems. Some towns i drive in are infuriating just based on lights. Then there's the growing fleets of all forms of deliveries through space 🤔🤔🤔important?

τony τao5 months ago

@StevenDiam77921 Very nice

Teodor6 months ago

@StevenDiam77921 @grok Wtf does that mean

Angry Davee5 months ago

@StevenDiam77921 What exactly are the opportunities on TIG for normal folks like me

Related Videos

This is why $TIG is so important 🔵 "We have got to acknowledge that most of the advances in AI came out of algorithm advances, not just the raw hardware" ⚫️ "The only way to really get 10x or 100x leaps is to fundamentally change the algorithm" Not just anyone's opinion.. thats from CEO of NVIDIA Jensen Huang Algorithms are where the real breakthroughs come from.. The Innovation Game (𝔦, 𝔦) is building the incentive layer to discover them.. at scale, in the open, with the best minds competing to push the frontier forward $TIG

This is why $TIG is so important 🔵 "We have got to acknowledge that most of the advances in AI came out of algorithm advances, not just the raw hardware" ⚫️ "The only way to really get 10x or 100x leaps is to fundamentally change the algorithm" Not just anyone's opinion.. thats from CEO of NVIDIA Jensen Huang Algorithms are where the real breakthroughs come from.. The Innovation Game (𝔦, 𝔦) is building the incentive layer to discover them.. at scale, in the open, with the best minds competing to push the frontier forward $TIG

𝗗𝗥𝗘𝗔𝗗 𝗕𝗢𝗡𝗚𝗢

21,675 views • 6 months ago

Yeah $TIG is flying under the radar An open protocol for algorithm development listening to John explain how algorithms can optimise & help you extract maximum utility from any given chip DeepMind, OpenAI, Anthropic, they're all racing to own the algorithmic layer if they win, you pay a tax on every optimized query, every solved route, every compressed model Over the last 50 years, we focused on hardware The fight was between Intel, ASML, TSMC: who could etch circuits smallest and cheapest at scale? many years ago we realised Moore's law seemed to be hitting a ceiling, that there may be fundamental limitations of physics as such, the focus changed to how we could extract maximum utility from the hardware the fight will be for the most optimised software I spoke with John Fletcher (𝔦, 𝔦) to discuss the TIG mission Full ep on our youtube 00:00 Intro 00:26 Primer on $TIG 02:18 Cambridge, ARM and the TIG Team 06:33 Why open source? 15:21 Why challenges for algorithms 20:00 Role of IP licensing 23:29 Defending Against Algorithmic Monopolies 27:40 Monetisation 28:40 Use Cases 36:23 Power Struggle vs AI Labs 39:12 Freedom Ideology

Yeah $TIG is flying under the radar An open protocol for algorithm development listening to John explain how algorithms can optimise & help you extract maximum utility from any given chip DeepMind, OpenAI, Anthropic, they're all racing to own the algorithmic layer if they win, you pay a tax on every optimized query, every solved route, every compressed model Over the last 50 years, we focused on hardware The fight was between Intel, ASML, TSMC: who could etch circuits smallest and cheapest at scale? many years ago we realised Moore's law seemed to be hitting a ceiling, that there may be fundamental limitations of physics as such, the focus changed to how we could extract maximum utility from the hardware the fight will be for the most optimised software I spoke with John Fletcher (𝔦, 𝔦) to discuss the TIG mission Full ep on our youtube 00:00 Intro 00:26 Primer on $TIG 02:18 Cambridge, ARM and the TIG Team 06:33 Why open source? 15:21 Why challenges for algorithms 20:00 Role of IP licensing 23:29 Defending Against Algorithmic Monopolies 27:40 Monetisation 28:40 Use Cases 36:23 Power Struggle vs AI Labs 39:12 Freedom Ideology

Jack Ai-Leung

40,766 views • 5 months ago

Introducing Axiom's discovery team: Alberto Alfarano Alberto's path runs from the IMO → trading → training AI to crack a 130-year-old conjecture. He found new Lyapunov functions while no general algorithm existed before. AI for math discovery redefines the boundary of human minds.

Introducing Axiom's discovery team: Alberto Alfarano Alberto's path runs from the IMO → trading → training AI to crack a 130-year-old conjecture. He found new Lyapunov functions while no general algorithm existed before. AI for math discovery redefines the boundary of human minds.

Carina Hong

373,882 views • 10 months ago

Our work in discovery is never done. Here’s the scoop on three key changes we made to our Recommended For You algorithm to connect more players with the games they’ll love.

Our work in discovery is never done. Here’s the scoop on three key changes we made to our Recommended For You algorithm to connect more players with the games they’ll love.

Roblox

119,405 views • 4 months ago

One of the most asked questions in TIG: how does the licensing work? Last week's community call answered it in detail. Arm Holdings put its chip designs in every phone on earth without manufacturing a single one. Licensing did that. The man who ran Arm's legal side is a TIG cofounder, and last week he explained how algorithms get sold for the first time in history.

One of the most asked questions in TIG: how does the licensing work? Last week's community call answered it in detail. Arm Holdings put its chip designs in every phone on earth without manufacturing a single one. Licensing did that. The man who ran Arm's legal side is a TIG cofounder, and last week he explained how algorithms get sold for the first time in history.

The Innovation Game (𝔦, 𝔦)

19,883 views • 3 months ago

Today Google DeepMind released AlphaEvolve: a Gemini coding agent for algorithm discovery. It beat the famous Strassen algorithm for matrix multiplication set 56 years ago. Google has been killing it recently. We had early access to the paper and interviewed the researchers.

Today Google DeepMind released AlphaEvolve: a Gemini coding agent for algorithm discovery. It beat the famous Strassen algorithm for matrix multiplication set 56 years ago. Google has been killing it recently. We had early access to the paper and interviewed the researchers.

Machine Learning Street Talk

58,517 views • 1 year ago

Introducing ALE-Bench, ALE-Agent! Towards Automating Long-Horizon Algorithm Engineering for Hard Optimization Problems Blog: Paper: ALE-Bench is a coding benchmark primarily focused on hard optimization (NP-hard) problems. We developed this benchmark with AtCoder Inc., a leading coding contest platform company. What makes ALE-Bench unique is its focus on hard optimization problems that demand long-horizon and creative reasoning. It’s open-ended, in the sense that true optima are out of reach (NP-hard) and scores can continuously improve. We believe this benchmark has the potential to become one of the key benchmarks for reasoning and coding in the next generation. ALE-Agent is our end-to-end agent that we specifically designed for this challenging domain. In fact, our ALE-Agent has already built an impressive track record in the wild! In May 2025, our agent participated in a live AtCoder Heuristic Competition (AHC), alongside 1,000 other participants in real-time. AHC is considered to be one of the most challenging coding competitions in this domain. Our ALE-Agent achieved an impressive ranking of 21st out of 1,000 human participants in the competition (top 2%), marking a turning point for AI discovery of solutions to hard optimization problems with a wide spectrum of important real world applications such as logistics, routing, packing, factory production planning, power-grid balancing. We look forward to applying this technology to real industrial optimization opportunities. Building on the insights from this study, Sakana AI will continue to tackle the challenge of developing AI with even greater algorithm engineering capabilities. ALE-Bench Dataset: ALE-Bench Code: This research was conducted in collaboration with AtCoder Inc. (AtCoder). We are deeply grateful for their outstanding expertise and contributions in optimization and algorithms, which were invaluable in providing data, analyzing results, and enabling our AI agent’s participation in their contests.

Introducing ALE-Bench, ALE-Agent! Towards Automating Long-Horizon Algorithm Engineering for Hard Optimization Problems Blog: Paper: ALE-Bench is a coding benchmark primarily focused on hard optimization (NP-hard) problems. We developed this benchmark with AtCoder Inc., a leading coding contest platform company. What makes ALE-Bench unique is its focus on hard optimization problems that demand long-horizon and creative reasoning. It’s open-ended, in the sense that true optima are out of reach (NP-hard) and scores can continuously improve. We believe this benchmark has the potential to become one of the key benchmarks for reasoning and coding in the next generation. ALE-Agent is our end-to-end agent that we specifically designed for this challenging domain. In fact, our ALE-Agent has already built an impressive track record in the wild! In May 2025, our agent participated in a live AtCoder Heuristic Competition (AHC), alongside 1,000 other participants in real-time. AHC is considered to be one of the most challenging coding competitions in this domain. Our ALE-Agent achieved an impressive ranking of 21st out of 1,000 human participants in the competition (top 2%), marking a turning point for AI discovery of solutions to hard optimization problems with a wide spectrum of important real world applications such as logistics, routing, packing, factory production planning, power-grid balancing. We look forward to applying this technology to real industrial optimization opportunities. Building on the insights from this study, Sakana AI will continue to tackle the challenge of developing AI with even greater algorithm engineering capabilities. ALE-Bench Dataset: ALE-Bench Code: This research was conducted in collaboration with AtCoder Inc. (AtCoder). We are deeply grateful for their outstanding expertise and contributions in optimization and algorithms, which were invaluable in providing data, analyzing results, and enabling our AI agent’s participation in their contests.

Sakana AI

239,187 views • 1 year ago

What evolves together fits together. We’re using RL to post-train models for Thermo AI research, accelerating algorithmic discovery for our new species of computer. The first sparks of Thermo RSI, in collaboration with Prime Intellect.

What evolves together fits together. We’re using RL to post-train models for Thermo AI research, accelerating algorithmic discovery for our new species of computer. The first sparks of Thermo RSI, in collaboration with Prime Intellect.

Extropic

159,863 views • 18 days ago

(1/3) insitro is excited to share our paper “EmbedGEM: A Framework to Evaluate the Utility of Embeddings for Genetic Discovery,” focusing on assessing the use of ML-derived embeddings for genetic discovery using heritability & relevance to specific diseases.

(1/3) insitro is excited to share our paper “EmbedGEM: A Framework to Evaluate the Utility of Embeddings for Genetic Discovery,” focusing on assessing the use of ML-derived embeddings for genetic discovery using heritability & relevance to specific diseases.

insitro

18,421 views • 2 years ago

THE INNOVATION GAME ($TIG) SURGES 100% IN 7 DAYS The Innovation Game (𝔦, 𝔦) ($TIG), a Base ecosystem token built around decentralized algorithmic research, is up roughly 101% over the past 7 days, making it one of the top performers in the Base Ecosystem and Proof of Work categories. TIG runs as a network for "algorithmic breakthroughs," using proof-of-work to incentivize open contributions across AI, cryptography, biomedical research, and climate science. The project formed a Swiss Association in Zug last year to safeguard high-value IP from global researchers and now sits at roughly $39M market cap with over 20,000 holders.

THE INNOVATION GAME ($TIG) SURGES 100% IN 7 DAYS The Innovation Game (𝔦, 𝔦) ($TIG), a Base ecosystem token built around decentralized algorithmic research, is up roughly 101% over the past 7 days, making it one of the top performers in the Base Ecosystem and Proof of Work categories. TIG runs as a network for "algorithmic breakthroughs," using proof-of-work to incentivize open contributions across AI, cryptography, biomedical research, and climate science. The project formed a Swiss Association in Zug last year to safeguard high-value IP from global researchers and now sits at roughly $39M market cap with over 20,000 holders.

BSCN

10,453 views • 5 months ago

Can LLMs invent better ways to train LLMs? At Sakana AI, we’re pioneering AI-driven methods to automate AI research and discovery. We’re excited to release DiscoPOP: a new SOTA preference optimization algorithm that was discovered and written by an LLM! Our method leverages LLMs to propose and implement new preference optimization algorithms. We then train models with those algorithms and evaluate their performance, providing feedback to the LLM. By repeating this process for multiple generations in an evolutionary loop, the LLM discovers many highly-performant and novel preference optimization objectives! Paper: GitHub: Model: We proudly collaborated with the University of Oxford (Foerster Lab for AI Research (now part of BOLD)) and Cambridge University (Mihaela van der Schaar) on this groundbreaking project. Looking ahead, we envision a future where AI-driven research reduces the need for extensive human intervention and computational resources. This will accelerate scientific discoveries and innovation, pushing the boundaries of what AI can achieve.

Can LLMs invent better ways to train LLMs? At Sakana AI, we’re pioneering AI-driven methods to automate AI research and discovery. We’re excited to release DiscoPOP: a new SOTA preference optimization algorithm that was discovered and written by an LLM! Our method leverages LLMs to propose and implement new preference optimization algorithms. We then train models with those algorithms and evaluate their performance, providing feedback to the LLM. By repeating this process for multiple generations in an evolutionary loop, the LLM discovers many highly-performant and novel preference optimization objectives! Paper: GitHub: Model: We proudly collaborated with the University of Oxford (Foerster Lab for AI Research (now part of BOLD)) and Cambridge University (Mihaela van der Schaar) on this groundbreaking project. Looking ahead, we envision a future where AI-driven research reduces the need for extensive human intervention and computational resources. This will accelerate scientific discoveries and innovation, pushing the boundaries of what AI can achieve.

Sakana AI

556,037 views • 2 years ago

Introducing the Cline SDK. We rebuilt the Cline harness for our extension and CLI from scratch using all the lessons learned since creating one of the world's first coding agents in 2024, and are open sourcing it for others to build with today. npm i @cline/sdk 🧵

Introducing the Cline SDK. We rebuilt the Cline harness for our extension and CLI from scratch using all the lessons learned since creating one of the world's first coding agents in 2024, and are open sourcing it for others to build with today. npm i @cline/sdk 🧵

Cline

3,581,677 views • 5 months ago

We built the first AI system to achieve the industry standard CAD certification for mechanical engineers. Our agent autonomously built parts and assembled mechanisms, passing the SOLIDWORKS Professional exam in less than 3 hours. Details below, including a concerning discovery.

We built the first AI system to achieve the industry standard CAD certification for mechanical engineers. Our agent autonomously built parts and assembled mechanisms, passing the SOLIDWORKS Professional exam in less than 3 hours. Details below, including a concerning discovery.

Spectral Labs

90,424 views • 5 days ago

We’re launching Mundi, the first open-source web GIS built for AI. After years of training geospatial AI models, we decided now is the moment to build the GIS software of the next decade. After the AI Vectorizer, Georeferencer, and Kue (our LLM agent inside QGIS), we realized that GIS is at an inflection point. But existing mapping software was neither powerful enough to allow for synchronous collaboration with AI agents, nor open source to allow the community to build on top of it. If we want to get the most out of LLMs, we need to create our own GIS that is purposefully designed for AI—Mundi. With Mundi, you collaborate with Kue, our AI GIS agent, to edit your map, rather than physically interacting with tools as you would in a legacy GIS.

We’re launching Mundi, the first open-source web GIS built for AI. After years of training geospatial AI models, we decided now is the moment to build the GIS software of the next decade. After the AI Vectorizer, Georeferencer, and Kue (our LLM agent inside QGIS), we realized that GIS is at an inflection point. But existing mapping software was neither powerful enough to allow for synchronous collaboration with AI agents, nor open source to allow the community to build on top of it. If we want to get the most out of LLMs, we need to create our own GIS that is purposefully designed for AI—Mundi. With Mundi, you collaborate with Kue, our AI GIS agent, to edit your map, rather than physically interacting with tools as you would in a legacy GIS.

Brendan Ashworth

41,230 views • 1 year ago

🗣️ Mikel Arteta on having key players available again: “The main focus has to be to improve our players and improve the connection between them because they are really, really good. “Then we can talk about attacking inputs. Last season we missed some of those players for a lot of months, and the team still had the ability to perform and win a lot of matches. “When we have them available, that’s a bonus.”

🗣️ Mikel Arteta on having key players available again: “The main focus has to be to improve our players and improve the connection between them because they are really, really good. “Then we can talk about attacking inputs. Last season we missed some of those players for a lot of months, and the team still had the ability to perform and win a lot of matches. “When we have them available, that’s a bonus.”

Connor Humm

17,823 views • 1 month ago

Generative AI is about to create the most addictive algorithms in human history. Spotify’s Co-President, Gustav Söderström, warns that because AI can understand you on a deeply personal level, the potential for digital addiction is massive. But AI is a dual-use technology. Instead of letting the algorithm dictate who you are, you can use it to program who you want to become. Here is the framework he breaks down: Implicit vs. Intentional: Current algorithms track what you *did*. AI allows you to explicitly tell it what you *want* to do. The English Interface: Historically, only a few million developers could dictate what computers did. Now, anyone can reprogram their digital experience using plain English. The future of tech isn't about letting AI control us it's about using AI to take back control of our own attention and habits.

Generative AI is about to create the most addictive algorithms in human history. Spotify’s Co-President, Gustav Söderström, warns that because AI can understand you on a deeply personal level, the potential for digital addiction is massive. But AI is a dual-use technology. Instead of letting the algorithm dictate who you are, you can use it to program who you want to become. Here is the framework he breaks down: Implicit vs. Intentional: Current algorithms track what you did. AI allows you to explicitly tell it what you want to do. The English Interface: Historically, only a few million developers could dictate what computers did. Now, anyone can reprogram their digital experience using plain English. The future of tech isn't about letting AI control us it's about using AI to take back control of our own attention and habits.

Block 🧱

31,383 views • 4 months ago

Palestine vs the Algorithm! Palestine versus the Algorithm is a fight happening every day. As the Israelis kill journalists, the Algorithm suppresses what they say. The Algorithm has been programmed to mute talk of Palestinian genocide; To gate keep the starvation and the slaughter, and the daily massacres to hide. Palestine versus the Algorithm is one of the great battles of our time. As the Algorithm supports mass murder, TikTok, Youtube, X, all fall in line. But the Algorithm is man-made and paid for by a Zionist agenda, To pervert social media platforms into becoming genocide defenders. Palestine versus the Algorithm, or Technology versus the People. The Algorithm is anti-human, blocking the good that would end the evil. It’s not only Palestinians that the Algorithm is designed to defeat, It's everyone who stands up for life and justice, and who refuses to retreat. Subscribe and Follow Mama on Rumble:

Palestine vs the Algorithm! Palestine versus the Algorithm is a fight happening every day. As the Israelis kill journalists, the Algorithm suppresses what they say. The Algorithm has been programmed to mute talk of Palestinian genocide; To gate keep the starvation and the slaughter, and the daily massacres to hide. Palestine versus the Algorithm is one of the great battles of our time. As the Algorithm supports mass murder, TikTok, Youtube, X, all fall in line. But the Algorithm is man-made and paid for by a Zionist agenda, To pervert social media platforms into becoming genocide defenders. Palestine versus the Algorithm, or Technology versus the People. The Algorithm is anti-human, blocking the good that would end the evil. It’s not only Palestinians that the Algorithm is designed to defeat, It's everyone who stands up for life and justice, and who refuses to retreat. Subscribe and Follow Mama on Rumble:

TheMamaSchwabParody

29,617 views • 1 year ago

The people who are making millions of dollars with A.I. aren’t smarter than you. They didn’t build the tools they’re using, they aren’t writing algorithms. They just saw all the new A.I. tools available and looked for ways to use them. They just tried while you did nothing.

The people who are making millions of dollars with A.I. aren’t smarter than you. They didn’t build the tools they’re using, they aren’t writing algorithms. They just saw all the new A.I. tools available and looked for ways to use them. They just tried while you did nothing.

Andrew Tate

1,009,591 views • 1 year ago

Linda McMahon on good uses of AI in schools for students: "They spend the first couple of hours with AI and then the rest of the day they are implementing things they have learned. It's remarkable."

Linda McMahon on good uses of AI in schools for students: "They spend the first couple of hours with AI and then the rest of the day they are implementing things they have learned. It's remarkable."

Aaron Rupar

908,971 views • 2 months ago