Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

1/4 LLMs solve research grade math problems but struggle with basic calculations. We bridge this gap by turning them to computers. We built a computer INSIDE a transformer that can run programs for millions of steps in seconds solving even the hardest Sudokus with 100% accuracy

Christos Tzamos

23,406 subscribers

1,823,781 görüntüleme • 5 ay önce •via X (Twitter)

Eğitim Bilim & Teknoloji

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

🇺🇸 NVIDIA CEO JENSEN HUANG: WE HAD TO INVENT THE COMPUTER BEFORE THE PROBLEM EXISTED Jensen reflected on Nvidia’s early days, revealing how the company’s mission in 1993 seemed almost impossible, building a computer for problems that didn’t yet exist. He said they had to “create the problem” before they could solve it, a gamble that shaped modern computing: “When Nvidia started in 1993, we were trying to create this new computing approach. The question is, what’s the killer app? The company wanted to create a new type of computer that could solve problems normal computers couldn’t. But the applications that existed in 1993 were ones that normal computers could solve. We had a mission statement for a company that had no chance of success. I didn’t know that in 1993. It just sounded like a good idea.” Source: Joe Rogan

🇺🇸 NVIDIA CEO JENSEN HUANG: WE HAD TO INVENT THE COMPUTER BEFORE THE PROBLEM EXISTED Jensen reflected on Nvidia’s early days, revealing how the company’s mission in 1993 seemed almost impossible, building a computer for problems that didn’t yet exist. He said they had to “create the problem” before they could solve it, a gamble that shaped modern computing: “When Nvidia started in 1993, we were trying to create this new computing approach. The question is, what’s the killer app? The company wanted to create a new type of computer that could solve problems normal computers couldn’t. But the applications that existed in 1993 were ones that normal computers could solve. We had a mission statement for a company that had no chance of success. I didn’t know that in 1993. It just sounded like a good idea.” Source: Joe Rogan

Mario Nawfal

19,268 görüntüleme • 8 ay önce

"What is P vs NP and how does it relate to human knowledge?" Avi Wigderson (Turing Award & Abel Prize winner): "In a very basic way, all the problems we want to solve are really NP problems. A subset of it is the problems we can solve. So the question of P versus NP is whether we can solve all the problems we want to solve, whether we can know everything we want to know. This is basically about the limits of our knowledge. You can think there is no limit to what we may want to know but I claim that essentially in any human endeavor, what people are searching for have the property that when a solution is found, it is easily recognized. NP are all problems that we can honestly say we really want to solve and P are those that we can solve. Whether they are equal is just the very fact that a solution can be easily recognized would imply that it can be efficiently found. That means that we can know everything we ever want to know, it's a fundamental question about human knowledge."

"What is P vs NP and how does it relate to human knowledge?" Avi Wigderson (Turing Award & Abel Prize winner): "In a very basic way, all the problems we want to solve are really NP problems. A subset of it is the problems we can solve. So the question of P versus NP is whether we can solve all the problems we want to solve, whether we can know everything we want to know. This is basically about the limits of our knowledge. You can think there is no limit to what we may want to know but I claim that essentially in any human endeavor, what people are searching for have the property that when a solution is found, it is easily recognized. NP are all problems that we can honestly say we really want to solve and P are those that we can solve. Whether they are equal is just the very fact that a solution can be easily recognized would imply that it can be efficiently found. That means that we can know everything we ever want to know, it's a fundamental question about human knowledge."

Ryan Peterman

20,069 görüntüleme • 2 ay önce

The verdict is in! Frontier models can pass the bar, yet they struggle on comprehensive legal research Today we're releasing Legal Research Bench, a benchmark that measures models’ ability to solve realistic legal research tasks across eight areas of U.S. law Instead of awarding partial credit, Legal Research Bench measures whether a model can conduct exhaustive legal analysis. We grade against a strict, all-pass rubric written by practicing lawyers. A model only receives full credit if every required legal element is correct Claude Opus 4.8 leads with 43.8% all-pass accuracy, followed by GPT 5.5 (40.4%) and Claude Sonnet 4.6 (38.5%). While top models score around 80% with partial credit, none exceed 44% when every required legal element must be correct The gap between partial and all-pass accuracy shows how difficult it remains for AI to produce complete, reliable legal research. We hope that Legal Research Bench helps better measure, and ultimately close that gap Lots of exciting work happening in Legal AI from Harvey and Crosby. Excited for the legal research benchmarks ahead!

The verdict is in! Frontier models can pass the bar, yet they struggle on comprehensive legal research Today we're releasing Legal Research Bench, a benchmark that measures models’ ability to solve realistic legal research tasks across eight areas of U.S. law Instead of awarding partial credit, Legal Research Bench measures whether a model can conduct exhaustive legal analysis. We grade against a strict, all-pass rubric written by practicing lawyers. A model only receives full credit if every required legal element is correct Claude Opus 4.8 leads with 43.8% all-pass accuracy, followed by GPT 5.5 (40.4%) and Claude Sonnet 4.6 (38.5%). While top models score around 80% with partial credit, none exceed 44% when every required legal element must be correct The gap between partial and all-pass accuracy shows how difficult it remains for AI to produce complete, reliable legal research. We hope that Legal Research Bench helps better measure, and ultimately close that gap Lots of exciting work happening in Legal AI from Harvey and Crosby. Excited for the legal research benchmarks ahead!

Vals AI

17,351 görüntüleme • 1 ay önce

This is the new campus of the Kyiv School of Economics. It will host engineering, math, computer science programs. We are building a world class university to make Ukraine European and prosperous. But we also focus on engineering so that Ukraine can defend itself

This is the new campus of the Kyiv School of Economics. It will host engineering, math, computer science programs. We are building a world class university to make Ukraine European and prosperous. But we also focus on engineering so that Ukraine can defend itself

Tymofiy Mylovanov

17,987 görüntüleme • 1 yıl önce

[Transformer] by Hand✍️📺 5-minute Video Tutorial Anna Rahn made this short video to explain the Transformer exercise for my Computer Vision course last spring. In 5 minutes, she demonstrates the key calculations of the Transformer by hand with pen and paper! Anna is a fantastic student. I am lucky to have her in my lab!

[Transformer] by Hand✍️📺 5-minute Video Tutorial Anna Rahn made this short video to explain the Transformer exercise for my Computer Vision course last spring. In 5 minutes, she demonstrates the key calculations of the Transformer by hand with pen and paper! Anna is a fantastic student. I am lucky to have her in my lab!

Tom Yeh

133,460 görüntüleme • 2 yıl önce

Axiom Math founder Carina Hong on why math is the path to general superintelligence: “Math is the sandbox for reality. You very quickly see verifiable rewards because in math, there’s an absolute right or wrong.” “And especially when you have Lean, you can check the proof or solution step by step. You’ll be able to apply reinforcement learning in a much more efficient way.” “We have currently scaled from winning a Putnam perfect score to solving a batch of research problems that professional mathematicians find really challenging. And we also see this transfer to code verification.”

Axiom Math founder Carina Hong on why math is the path to general superintelligence: “Math is the sandbox for reality. You very quickly see verifiable rewards because in math, there’s an absolute right or wrong.” “And especially when you have Lean, you can check the proof or solution step by step. You’ll be able to apply reinforcement learning in a much more efficient way.” “We have currently scaled from winning a Putnam perfect score to solving a batch of research problems that professional mathematicians find really challenging. And we also see this transfer to code verification.”

TBPN

39,589 görüntüleme • 4 ay önce

QUANTUM COMPUTING: THE FUTURE IS NOW IBM recently announced a quantum computer capable of solving problems in minutes that would take today's supercomputers millions of years. Developed by IBM, Google, and others, quantum computers push the boundaries of science and engineering, utilizing atomic forces to achieve unprecedented computational power. A simple maze shows how quantum computing can outpace classic computers. Michio Kaku: “All the big boys who are in this race to create a workable, operationally efficient quantum computer because the nation or company that does this will rule the world economy.” Source: 60 Minutes

QUANTUM COMPUTING: THE FUTURE IS NOW IBM recently announced a quantum computer capable of solving problems in minutes that would take today's supercomputers millions of years. Developed by IBM, Google, and others, quantum computers push the boundaries of science and engineering, utilizing atomic forces to achieve unprecedented computational power. A simple maze shows how quantum computing can outpace classic computers. Michio Kaku: “All the big boys who are in this race to create a workable, operationally efficient quantum computer because the nation or company that does this will rule the world economy.” Source: 60 Minutes

Mario Nawfal

327,898 görüntüleme • 2 yıl önce

Milestone! We (robotic arms for gadgets assembly) finished the first commercial order, which brought the first revenue. Here are some learnings from this: The customer was a smart toy manufacturer. The task was to add a heatsink to Raspberry Pi. We received parts from them and returned the assembled modules back. Currently, it's done by teleoperation. Later it will be done by a remote employee via the Internet. Then it will be automated action by action, reducing the operator's time on this and making the task profitable. ps. If you have an assembly task that we can do for you asynchronically - leave a comment below. Learning 1. It's possible! This task which is usually done by the human arm with 5 fingers can be done with a two-finger gripper with the addition of a couple of simple tooling. The task was not simplified. We peeled off thin films from stickers, unpacked paper boxes, moved PCB boards full of components, etc. And no unsolvable problems have been encountered yet. Challenges: 1) The paper box shifted during the opening Solved with the plastic walls that you can lean against 2) Heat pad, stuck to the gripper instead of heat sync. Can be solved by gripper with a pump, but this time solved with the patience of the operator 3) The film on the pad is very thin. Turned out that sub-millimeter arm precision is enough to peel it off with just a regular gripper. 4) The working area has not enough space. You'll only know this by doing real tasks in bulk. This could be solved by an extra pair of long arms, but in this case, solved with the patience of the operator. I think that in the end, we will have 5-10 types of universal tooling and 5-10 types of grippers to solve almost all the problems in such assembly tasks. Learning 2. It's slow. It took 5 times more time, than doing it with human hands. But the good news is there's a lot of room for improvement. We now have specific “time for task” metrics, which we will decrease with iterations. The main reasons for slowness: 1) To rotate the gripper to a steep angle you are forced to control one robot arm with two hands instead of using both arms. We can fix this by just making more room for rotations. 2) Grabbing PCB board with two arms is hard. A slight difference in rotation can break the board, and it's hard to control these angles visually. To solve this, the best way is to use force feedback so you can feel the pressure applied to the item. 3) Accuracy and steadiness is still can be improved We will try a metal version and double the motors to do this. 4) It is physically difficult for the human hands to move with such precision To solve this, we will add a pad for the hands like in surgical robots Learning 3. It's a good business model The "Factory in the cloud" is a good business model for this stage. You send us parts and we send back assembled modules. Currently, it's more convenient than sending a robot to your place, as we can iterate/fix the robot quickly and utilize it 100% of the time. When we polish the set-up over time - we can send robots to your place. So if we can assemble something for you in the USA with Chinese prices by using modern automation - leave a comment below.

Milestone! We (robotic arms for gadgets assembly) finished the first commercial order, which brought the first revenue. Here are some learnings from this: The customer was a smart toy manufacturer. The task was to add a heatsink to Raspberry Pi. We received parts from them and returned the assembled modules back. Currently, it's done by teleoperation. Later it will be done by a remote employee via the Internet. Then it will be automated action by action, reducing the operator's time on this and making the task profitable. ps. If you have an assembly task that we can do for you asynchronically - leave a comment below. Learning 1. It's possible! This task which is usually done by the human arm with 5 fingers can be done with a two-finger gripper with the addition of a couple of simple tooling. The task was not simplified. We peeled off thin films from stickers, unpacked paper boxes, moved PCB boards full of components, etc. And no unsolvable problems have been encountered yet. Challenges: 1) The paper box shifted during the opening Solved with the plastic walls that you can lean against 2) Heat pad, stuck to the gripper instead of heat sync. Can be solved by gripper with a pump, but this time solved with the patience of the operator 3) The film on the pad is very thin. Turned out that sub-millimeter arm precision is enough to peel it off with just a regular gripper. 4) The working area has not enough space. You'll only know this by doing real tasks in bulk. This could be solved by an extra pair of long arms, but in this case, solved with the patience of the operator. I think that in the end, we will have 5-10 types of universal tooling and 5-10 types of grippers to solve almost all the problems in such assembly tasks. Learning 2. It's slow. It took 5 times more time, than doing it with human hands. But the good news is there's a lot of room for improvement. We now have specific “time for task” metrics, which we will decrease with iterations. The main reasons for slowness: 1) To rotate the gripper to a steep angle you are forced to control one robot arm with two hands instead of using both arms. We can fix this by just making more room for rotations. 2) Grabbing PCB board with two arms is hard. A slight difference in rotation can break the board, and it's hard to control these angles visually. To solve this, the best way is to use force feedback so you can feel the pressure applied to the item. 3) Accuracy and steadiness is still can be improved We will try a metal version and double the motors to do this. 4) It is physically difficult for the human hands to move with such precision To solve this, we will add a pad for the hands like in surgical robots Learning 3. It's a good business model The "Factory in the cloud" is a good business model for this stage. You send us parts and we send back assembled modules. Currently, it's more convenient than sending a robot to your place, as we can iterate/fix the robot quickly and utilize it 100% of the time. When we polish the set-up over time - we can send robots to your place. So if we can assemble something for you in the USA with Chinese prices by using modern automation - leave a comment below.

Igor Kulakov

37,266 görüntüleme • 1 yıl önce

In this simple example, we show that Claude Code can read the iMessage database on the latest version of macOS, even with a leading EDR running on the system, illustrating the impact of an adversary who can remotely control the agent. We do this using Terminator, an internal research tool we built while studying the security implications of computer use agents. In this setup, the terminal application has previously been granted FDA, a subtle misconfiguration that effectively gives the agent access to unexpected context.

In this simple example, we show that Claude Code can read the iMessage database on the latest version of macOS, even with a leading EDR running on the system, illustrating the impact of an adversary who can remotely control the agent. We do this using Terminator, an internal research tool we built while studying the security implications of computer use agents. In this setup, the terminal application has previously been granted FDA, a subtle misconfiguration that effectively gives the agent access to unexpected context.

Origin

18,004 görüntüleme • 8 ay önce

Deepa Soni, the executive vice president and CIO of New York Life Insurance Company sees AI as a "strategic enabler." “It’s not technology waiting to solve a business problem, but there are actual business problems that we can solve now with AI that we could not before.”

Deepa Soni, the executive vice president and CIO of New York Life Insurance Company sees AI as a "strategic enabler." “It’s not technology waiting to solve a business problem, but there are actual business problems that we can solve now with AI that we could not before.”

TIME

909,576 görüntüleme • 2 ay önce

The list of all the problems in mathematics that Terence Tao *could* solve is absolutely enormous. The problem is we can't make copies of him, dump millions of dollars of inference compute into each copy, and have them each tackle a different Millennium Prize Problem for a hundred subjective years. But we can do that with AI. So in that way, we may be underrating how significant it will be when we have AIs that are not superintelligent but rather simply as good as the best humans today. It’s actually an extremely powerful property of AI that by just running copies in parallel, you can immediately fill in the waterline of accomplishments that are feasible with a given level of capabilities.

The list of all the problems in mathematics that Terence Tao could solve is absolutely enormous. The problem is we can't make copies of him, dump millions of dollars of inference compute into each copy, and have them each tackle a different Millennium Prize Problem for a hundred subjective years. But we can do that with AI. So in that way, we may be underrating how significant it will be when we have AIs that are not superintelligent but rather simply as good as the best humans today. It’s actually an extremely powerful property of AI that by just running copies in parallel, you can immediately fill in the waterline of accomplishments that are feasible with a given level of capabilities.

Dwarkesh Patel

99,378 görüntüleme • 4 ay önce

Dr. Phil Goes Off Script With Stunning Criticism of the US Government “We got people in charge of solving problems that don’t really want to solve them because their job depends on having that problem. “They need that problem to justify their existence. They need that problem to maintain this huge bureaucracy, to maintain this big budget, this big agency. They need that problem to exist. I think we’ve got people in charge of solving problems that are not problem solvers.”

Dr. Phil Goes Off Script With Stunning Criticism of the US Government “We got people in charge of solving problems that don’t really want to solve them because their job depends on having that problem. “They need that problem to justify their existence. They need that problem to maintain this huge bureaucracy, to maintain this big budget, this big agency. They need that problem to exist. I think we’ve got people in charge of solving problems that are not problem solvers.”

The Vigilant Fox 🦊

2,147,282 görüntüleme • 2 yıl önce

As a first step, we are launching NaijaCivicTech, an open source platform where Nigerian in tech curate and build tiny tools that solve specific Nigerian problems together this is not a startup, not a pitch deck, just working software in the open, built for Nigeria for ease. The current goal is to achieve two things: 1. Community to curate all existing tools that have attempted to solve a problem but are not well known. 2. Community to suggest problems, tools and project that solves a specific problem It's day 1, let's see how far we can go.

As a first step, we are launching NaijaCivicTech, an open source platform where Nigerian in tech curate and build tiny tools that solve specific Nigerian problems together this is not a startup, not a pitch deck, just working software in the open, built for Nigeria for ease. The current goal is to achieve two things: 1. Community to curate all existing tools that have attempted to solve a problem but are not well known. 2. Community to suggest problems, tools and project that solves a specific problem It's day 1, let's see how far we can go.

U N C L E BIGBAY ✨

33,813 görüntüleme • 4 ay önce

.Naval: Every human is a lottery ticket bet on the future of the species. One of the things that you really learn when you read David Deutsch’s theories and you authenticate them for yourself is you realize humans are universal explainers. That means everything that we know in the universe follows the laws of physics, and there’s no reason to believe otherwise. If you think otherwise, then please present your better theory that explains the world. If you can’t do that, then you have to go with the laws of physics. Well, the laws of physics are completely computable. They can fit inside a Turing machine or computer, and a computer can simulate the laws of physics with arbitrary accuracy, limited only by the specific power of that computer. If you increase the power of that computer, you can simulate them more accurately. So humans already simulate—in our minds we simulate—and through our computers we simulate the weather, we simulate quasars, we even simulate human systems. We simulate the economy. We simulate all kinds of things. So anything that can be understood, we can understand in our minds. This is something the AGI people get wrong when they talk about superintelligence. There is nothing out there that can understand something fundamentally that we can’t understand. It might be faster at it, it might have more compute, it might have more memory, but there’s no concept that it can understand that we can’t ourselves understand. So we are maximal universal explainers. That means every human is capable of unbounded creativity. Anyone could be the next Einstein or Fermi or Elon Musk or Jeff Bezos or Jonas Salk or whatever. So we can create anything. And if we can create anything, every human is a lottery ticket bet on the future of the species.

.Naval: Every human is a lottery ticket bet on the future of the species. One of the things that you really learn when you read David Deutsch’s theories and you authenticate them for yourself is you realize humans are universal explainers. That means everything that we know in the universe follows the laws of physics, and there’s no reason to believe otherwise. If you think otherwise, then please present your better theory that explains the world. If you can’t do that, then you have to go with the laws of physics. Well, the laws of physics are completely computable. They can fit inside a Turing machine or computer, and a computer can simulate the laws of physics with arbitrary accuracy, limited only by the specific power of that computer. If you increase the power of that computer, you can simulate them more accurately. So humans already simulate—in our minds we simulate—and through our computers we simulate the weather, we simulate quasars, we even simulate human systems. We simulate the economy. We simulate all kinds of things. So anything that can be understood, we can understand in our minds. This is something the AGI people get wrong when they talk about superintelligence. There is nothing out there that can understand something fundamentally that we can’t understand. It might be faster at it, it might have more compute, it might have more memory, but there’s no concept that it can understand that we can’t ourselves understand. So we are maximal universal explainers. That means every human is capable of unbounded creativity. Anyone could be the next Einstein or Fermi or Elon Musk or Jeff Bezos or Jonas Salk or whatever. So we can create anything. And if we can create anything, every human is a lottery ticket bet on the future of the species.

Arjun Khemani

33,110 görüntüleme • 1 yıl önce

How far can a very simple eye go in solving vision tasks? Like a 1-pixel camera? Humans have one of the greatest eyes in nature, while many animals have significantly simpler eyes and visual systems yet show complex perceptual behavior. In an interesting project, we find that many computer vision tasks can be solved without a typical camera and with such simple 1-pixel sensors (photoreceptors). We also find that proper design (e.g., where to place the photoreceptors strategically) makes a big difference, so we developed a computational design method to find them. 🌐 👁️[Solving Vision Tasks with Simple Photoreceptors Instead of Cameras] 🧵1/n

How far can a very simple eye go in solving vision tasks? Like a 1-pixel camera? Humans have one of the greatest eyes in nature, while many animals have significantly simpler eyes and visual systems yet show complex perceptual behavior. In an interesting project, we find that many computer vision tasks can be solved without a typical camera and with such simple 1-pixel sensors (photoreceptors). We also find that proper design (e.g., where to place the photoreceptors strategically) makes a big difference, so we developed a computational design method to find them. 🌐 👁️[Solving Vision Tasks with Simple Photoreceptors Instead of Cameras] 🧵1/n

Amir Zamir

75,892 görüntüleme • 2 yıl önce

Announcing Endex Deep Research in partnership with OpenAI. Powered by the OpenAI Deep Research API and our purpose-built indexing across millions of pages of financial documents, Endex agents excel at a variety of tasks. We have teamed up with Brett Caughran to showcase in action

Announcing Endex Deep Research in partnership with OpenAI. Powered by the OpenAI Deep Research API and our purpose-built indexing across millions of pages of financial documents, Endex agents excel at a variety of tasks. We have teamed up with Brett Caughran to showcase in action

Endex

68,012 görüntüleme • 1 yıl önce

We founded Obvious Ventures to invest in companies solving humanity's biggest challenges. In this year’s World Positive Report, we’re sharing with you 25 companies that are solving these problems, one piece of the puzzle at a time. #worldpositive

We founded Obvious Ventures to invest in companies solving humanity's biggest challenges. In this year’s World Positive Report, we’re sharing with you 25 companies that are solving these problems, one piece of the puzzle at a time. #worldpositive

Obvious Ventures

18,385 görüntüleme • 2 yıl önce

Baylor HC Dave Aranda - Defensive Philosophy - Selection Not Compression - Change the Math - Shrink the Run Area "Keep it simple enough so the player can grow. The player can take over the game." "It's a game of 11v11, but the person with the most at the point of attack is going to win. We can do that with fundamentals, but we can also do that with scheme."

Baylor HC Dave Aranda - Defensive Philosophy - Selection Not Compression - Change the Math - Shrink the Run Area "Keep it simple enough so the player can grow. The player can take over the game." "It's a game of 11v11, but the person with the most at the point of attack is going to win. We can do that with fundamentals, but we can also do that with scheme."

James Light

28,931 görüntüleme • 1 yıl önce

$*New Paper on AI & Democracy* Imagine two approaches to democracy. The one we have today, where citizens choose a professional politician to represent them and others. Or an augmented form of democracy, where each citizen controls a personalized AI that helps them participate in thousands of nuanced decisions. This second approach is the idea of Augmented Democracy I introduced six years ago at TED. In our latest paper we explore a simplified version of Augmented Democracy by combining off-the-shelf LLMs, such as ChatGPT, with data collected using a collaborative government program builder. This was an online game where people build a personalized government program using proposals extracted from the programs of the candidates of the 2022 presidential election in Brazil. So how accurate are these augmented forms of democracy? Imagine a user who gave us 40 answers. We can use the first 20 to fine-tune a model that we can test using the 20 answers the model didn’t see. We can then compare the accuracy of these predictions with the ones obtained by a “bundle” rule, which assumes that users that self-reported to be from the left or right always chose the proposals from the candidate that shares their political identity. This showed us that LLMs were more accurate at predicting policy preferences than the bundle rule, meaning that the preferences captured in the participation data were more nuanced than a left-right axis, and that the LLMs can capture some of that nuance. Also, the LLMs can choose among policies coming from the same candidate, which is something that we cannot do using a bundle rule. But can these LLMs help us complete the aggregate preferences of the population? Direct or unbundled forms of participation can result in incomplete data when people answer only a fraction of all questions. In our paper, we simulate this incompleteness by sampling the full dataset. We ask how close we can get to the full dataset by using a random sample, or a random sample augmented by predictions made by these LLMs. Overall, we find that LLM-augmented data gets much closer to the full dataset than a pure random sample. These results do not mean that augmented democracy technology is ready, but they means we are in a much better place to continue exploring this idea than six years ago. This paper was a collaborative effort with Jairo Gudino, PhD student at CCL at the University of Toulouse Capitole and Umberto Grandi from IRIT also at the University of Toulouse Capitole. We hope you find these results insightful!$

New Paper on AI & Democracy Imagine two approaches to democracy. The one we have today, where citizens choose a professional politician to represent them and others. Or an augmented form of democracy, where each citizen controls a personalized AI that helps them participate in thousands of nuanced decisions. This second approach is the idea of Augmented Democracy I introduced six years ago at TED. In our latest paper we explore a simplified version of Augmented Democracy by combining off-the-shelf LLMs, such as ChatGPT, with data collected using a collaborative government program builder. This was an online game where people build a personalized government program using proposals extracted from the programs of the candidates of the 2022 presidential election in Brazil. So how accurate are these augmented forms of democracy? Imagine a user who gave us 40 answers. We can use the first 20 to fine-tune a model that we can test using the 20 answers the model didn’t see. We can then compare the accuracy of these predictions with the ones obtained by a “bundle” rule, which assumes that users that self-reported to be from the left or right always chose the proposals from the candidate that shares their political identity. This showed us that LLMs were more accurate at predicting policy preferences than the bundle rule, meaning that the preferences captured in the participation data were more nuanced than a left-right axis, and that the LLMs can capture some of that nuance. Also, the LLMs can choose among policies coming from the same candidate, which is something that we cannot do using a bundle rule. But can these LLMs help us complete the aggregate preferences of the population? Direct or unbundled forms of participation can result in incomplete data when people answer only a fraction of all questions. In our paper, we simulate this incompleteness by sampling the full dataset. We ask how close we can get to the full dataset by using a random sample, or a random sample augmented by predictions made by these LLMs. Overall, we find that LLM-augmented data gets much closer to the full dataset than a pure random sample. These results do not mean that augmented democracy technology is ready, but they means we are in a much better place to continue exploring this idea than six years ago. This paper was a collaborative effort with Jairo Gudino, PhD student at CCL at the University of Toulouse Capitole and Umberto Grandi from IRIT also at the University of Toulouse Capitole. We hope you find these results insightful!

César A. Hidalgo

26,915 görüntüleme • 1 yıl önce