Archive for the ‘Alphazero’ Category

Creating New Stories That Don’t Suck – Hollywood in Toto

I was four years old when my father took me to the theater to see Star Wars.

From that moment, until that calamitous day in 1999 when The Phantom Menace debuted, the Star Wars trilogy was everything storytelling should be.

In the past few decades, stories like the ones I grew up loving have been in short supply, leaving me to ask:

And, of course

Hollywood in Toto helped to introduce me to The Critical Drinker. It was the Drinker who reverse-engineered the cultural and political rules that govern modern storytelling and why those rules result in terrible movies.

After watching a few dozen Drinker videos, I was convinced that I could create an epic adventure better than anything Hollywood is churning out these days: one that ignores the rules supposedly demanded by modern audiences and the mobs of fanatics on social media.

Further, I was determined to create something special for a very ordinary reason: I wanted to impress a girl. My creation had to be good because the girl I needed to impress is a very special one: my wife of 25 years.

That leads me to my concept

Imagine a computer algorithm that can know your hopes, dreams and deepest desires better than even your friends and family do. It can use that knowledge to persuade you to do whatever its owner wants you to do.

The possibility that such a technology could one day exist seemed terrifyingly plausible, given that tech giants sit atop enormous mountains of data about every one of us. Further, with all of the devices in our midst, they can spy on us and communicate with us in a multitude of ways.

Having learned something about the psychology of influence, I found it easy to think of persuasion as a rule-bound endeavor, much like the game of chess or the Japanese game called Go. Long before the world stood in amazement at the capabilities of ChatGPT, I had become deeply unsettled by the abilities of an AI program called AlphaZero that had revolutionized both chess and Go (mastery of Go had eluded computer science far longer than chess).

What if AI could master persuasion the way it has mastered these games? What use would it be put to? The first use seemed obvious. It would be used for the same purpose as any new technology: to get girls. But what then? The results amazed and terrified me.

I began to imagine a tech billionaire who owns an algorithm that can persuade and manipulate people more effectively than any human being can. Algorithm in hand, Neville first uses his program to seduce Meghan Peters, a Hollywood starlet whom he could never attract on his own.

Nevilles problem is that he owes the Chinese a lot of money. They come to him with their concerns that his program, while effective, is not perfect. Theyre tired of having to imprison dissidents for political crimes. It reflects badly on China on the world stage. They threaten to collect on their debts and ruin Neville if he doesnt perfect his program.

Neville is unable to meet their demand. The sheer complexity of the world makes the task of perfecting the program as he originally conceived it impossible. He and his team find a clever but terrifying workaround, one that solves the problem in the most horrifying way imaginable. His new weapon of persuasion interests both Chinese and American politicians.

To make Nevilles evil plan terrifyingly plausible, I had to take the reader through it one devious step at a time. If the reader doesnt see whats coming, it has grown organically out of the story. But if the reader begins to sense whats coming, the eventual payoff is even more terrifying.

To accomplish that, I realized I would have to out-Crichton Michael Crichton by filling my budding techno-thriller with lots of real-world techno: physics, mathematics, complex systems, psychology, computer science, and network science.

Additionally, having spent the past decade on a program of self-education, I was also able to fill the work with law, art, history, opera, and political intrigue, all of it working to enhance the plot and deepen the characters.

What kind of heroes could overcome the plans of an AI-backed genius whose work is by nature secret?

Realistic but positive and inspiring, I designed a cast of the most American of heroes: ordinary and imperfect characters of every background who have, (in varying degrees) some of the four cardinal virtues of prudence, justice, fortitude, and temperance.

Not knowing exactly what theyre up against, they all get in over their heads and they have to find their way out. Only one of my male characters is an incompetent laughingstock. Without having to cater to modern audiences I was able to build female characters who are genuinely strong rather than the strong female character we see so much on screen.

The men in Kingmaker are the kind you would recognize from your everyday life rather than the hyperactive hyper-emotional hyper-talkative children forced into mens bodies that todays insecure, effeminate, emotionally fragile Hollywood screenwriters imagine men to be.

The heroes in Kingmaker arent trying to establish some unreachable Utopia. They dont strive to be empowered because the desire to control the lives of others hardly seems like a noble calling fit for a hero.

Instead, theyre concerned with the same things we all are concerned with: finding work, finding love, finding meaning, finding forgiveness, exposing injustice. The obstacles they face are the same obstacles we all face: constant lying from our betters, utopian educators alienating our children with indoctrination, and the sheer amount of effort it takes to truly grow and achieve anything worthwhile.

In their pursuit, they show us all what we can become with courage, conviction, humility, and effort. In their quest, they uphold the best of our nations legacy and urge us to defend it from those who would tear it down.

As a writer, I wanted to experience the world from a multitude of perspectives by creating characters who are different from me. Some of my heroes are different from me in the trivial dimension of race. But theyre also different on the more important dimensions of experience, profession, and worldview.

For instance, although Im an orthodox Jew, one of my heroes is a Catholic priest.

After reading about Chinese spy Christine Fang who seduced Congressman Eric Swalwell and other American politicians, I just had to have a honeypot spy of my own to spice up the plot. The result was Mei Hua Chang, a dangerous wild card in the plot of Kingmaker. Shes a femme fatale whose sex appeal is only exceeded by her cunning.

To execute her character properly, I could only get her clothes off but once. In all of her other interactions, shes required to play the cards shes dealt to perfection. I always relished the challenge of writing her scenes.

WWCDD: What Would Critical Drinker Do?

By the time I finished planning the novel out, I had something like five plots going simultaneously. It reminded me of the HBO series Game of Thrones (but unlike George RR Martin, I actually finished writing the book, and unlike the HBO series, I provided an ending that works).

In order to tame all those plots into a coherent whole, I made a careful study of my characters, their backgrounds and their emotions. I attended to the practicalities of what my characters were attempting to do.

Inspired by Critical Drinker, I made a careful study of setup and payoff. Every plot point either had to set up an important payoff later or be the payoff of an earlier setup. (My daughter Leah nicknamed the book Chekhovs Arsenal.) That one Critical Drinker video became the inspiration for an entire method of writing that guided the whole process.

A writer could do much worse than to ask himself, What would the Drinker do?

Critical Drinker observes that modern screenwriters, bent on replacing legacy characters with their mediocre creations, seem to hate their fans. I was raised in a different generation. I drew on the teaching of Dale Carnegie.

He said of the great American magician Howard Thurston, that he would remind himself before every performance how much he loved his audience. I did something very similar before sitting down to write. That was easy because I was writing to impress the love of my life.

More than that. I began sending pieces of Kingmaker to my friend Martin. He devoured them and asked for more. Energized by his encouragement, I was able to keep on writing even when the going got difficult.

Critical Drinker once commented: One of the most disgusting hallmarks of modern screenwriting is the denigration of the past in a desperate attempt to elevate the present. The bastardization of other peoples work to service your own.

Kingmaker skewers all of those who would deconstruct the arts, architecture, and legacy popular culture IPs, revealing them to be dangerous, power-hungry operatives. It does so in ways that spring organically from and contribute to the plot.

For instance, when it dawns on my two main heroes that their favorite comic book movie series has lost interest in telling the great stories they once told, that realization dawns on each of them at different times. That difference moves the plot forward.

While Critical Drinker takes to task what the custodians of popular culture have done to the IPs they have been entrusted with, others have pointed out that the same thing is happening to high culture as well. Kingmaker addresses that and imagines what the next step will be for the arts if those who would deconstruct them get their way.

If a technology like Nevilles existed, which political party would be the one to use it?

The answer is obvious: either of them. Any political party can have ruthless operatives, opportunists, and time-serving apparatchiks. It should be obvious to every American that one party isnt the domain of well-meaning idealists while the other is the one for evil wannabe tyrants. If Machiavelli has taught us anything hes taught us that the public would never know which leader is virtuous and which one is simply an effective Machiavellian. Party affiliation cant help the public tell the good politicians from the bad.

While Kingmaker addresses issues such as lawfare and election integrity, which is currently associated with the political right, that issue is genuinely critical to all Americans. Governments that are secure enough in their power that they dont have to answer to the people have done some tremendously depraved things to keep and increase that power.

In those one-party states such as Saddams Iraq, it was extremely dangerous to be a member of the Baath party.

When offered Nevilles tool to increase their power, the villains in Kingmaker do not hesitate to add members of their own party to the proscription list.

In any political system, the strategy for getting and keeping power is the same: reward the people who keep you in power and punish those who oppose you. That goes for a dictator who depends on a small number of people to keep him in power and a Republic that (hopefully) governs with the consent of the people at large. The evil Party operatives in Kingmaker are not evil because of which party they choose. Theyre evil because as Machiavelli and Game Theory teach us, good politics frequently requires evil behavior.

Unlike modern moviemakers, I dont dictate my answers to complex questions to my readers. In fact, I have no idea how to answer many of the thorny questions posed by Kingmaker. For instance, what is consent in an age of powerful persuasion?

How do we make it so that the government truly can be said to have the consent of the governed? Did Meghan Peters genuinely consent to Jerry Nevilles advances after she had been seduced by his computer algorithm? I invite the reader to think through questions like these about the topic of consent with me. I dont attempt to answer them, even in a book entitled Consent.

One reason I dont attempt an answer is an important truth that Jerry Neville knows about the human mind: our brains are not sophisticated enough to genuinely understand a board game any more complicated than checkers. Equipped with that kind of brain, I realize that were going to have to grope our way through these issues and learn from experience rather than trusting some smarty-pants who says hes figured it all out for us.

I didnt want anything about Kingmaker to be dictated to me by a publishing industry that is busy editing out the offensive bits of James and the Giant Peach. Instead, I followed my vision.

Paul Joseph Watson has famously argued that populism is the new punk. Every step of this project has been influenced by the 1970s punk ethos of DIY or Die. I immersed myself in the nitty-gritty of every stage of this production: background learning, writing, editing, formatting, and the recording and editing of the audiobook.

The only thing I didnt do was design the cover. That was left to my friend Richard Smotherman. The result was a product I was so confident about that I put Consent, the first volume of Kingmaker on YouTube for free.

Facing down the impossible odds

If the culture war against those trying to destroy our heritage is to be won, we have to become creators. Its a risky business. The odds against hitting it big are long. The thought of daring greatly and having the world reject your work is not one that everybody can bear. Still, it must be done.

Simply finishing Kingmaker has been a reward unto itself. For the past two years, I have gotten to experience the greatest drama I will ever encounter. I started with a simple premise and thought as deeply as I could about what I had created.

Thinking deeply about simple things leads to wonderful discoveries. I was constantly delighted and surprised by what my characters did and the way the plot twisted and turned as I tried to wrestle five stories into a coherent whole. All the while, the project was spiced by the persistent question I asked of myself: Can you do this?

For two years, I willed that answer to be yes. No matter what, for the rest of my life, I will be able to say that I stared down those impossible odds and kept going when I thought I couldnt. In the process, I created something wonderful.

I hope you will take that risk and join me in creating culture. Our nation needs it. You might even impress your wife.

Attorney and polymath Ari H. Mendelson is the author of the Kingmaker trilogy. His previous novel was Bias Incident: The Worlds Most Politically Incorrect Novel. Before beginning the Kingmaker Trilogy, Mendelson dedicated himself to home schooling his four children. You can find his books at Amazon.com and GoodReads. Follow him on Twitter (X) via @kingmakerseries.

Go here to see the original:
Creating New Stories That Don't Suck - Hollywood in Toto

AI Agents: Adapting to the Future of Software Development – ReadWrite

In the near future, AI agents like Pixie from GPTConsole, Codeinterpreters from OpenAI, and many others are poised to revolutionize the software development landscape. They promise to supercharge mundane coding tasks and even autonomously build full-fledged software frameworks. However, their advanced capabilities bring into question the future role and relevance of human developers.

As these AI agents continue to proliferate, their efficiency and speed could potentially diminish the unique value human developers bring to the table. The rapid rise of AI in coding could alter not just the day-to-day tasks of developers but also have long-term implications for job markets and educational systems that prepare individuals for tech roles. Nick Bostrom raises two key challenges with AI.

The first, called the Orthogonality Thesis, suggests that an AI can be very smart but not necessarily share human goals. The second, known as the Value Loading Problem, highlights how difficult it is to teach an AI to have human values. Both these ideas feed into a more significant issue, the Problem of Control, which concerns the challenges of keeping these increasingly smart AIs under human control.

If not properly guided, these AI agents could operate in ways that are misaligned with human objectives or ethics. These concerns magnify the existing difficulties in effectively directing such powerful entities.

Despite these challenges, the incessant launch of new AI agents offers an unexpected silver lining. Human software developers now face a compelling need to elevate their skillsets and innovate like never before. In a world where AI agents are rolled out by the thousands daily, the emphasis on humans shifts towards attributes that AI cant replicatesuch as creative problem-solving, ethical considerations, and a nuanced understanding of human needs.

Rather than viewing the rise of AI as a threat, this could be a seminal moment for human ingenuity to flourish. By focusing on our unique human strengths, we might not just coexist with AI but synergistically collaborate to create a future that amplifies the best of both worlds. This sense of urgency is heightened by the exponential growth in technology, captured by Ray Kurzweils Law of Accelerating Returns.

The Law of Accelerating Returns by Ray Kurzweil intensifies the urgency, indicating that AI advancements will not only continue but accelerate, drastically shortening our time to adapt and innovate. The idea is simple: advancements arent linear, but accelerate over time.

For instance, simple life forms took billions of years to evolve into complex ones, but only a fraction of that time to go from complex forms to humanoids. This principle extends to cultural and technological changes, like the speed at which we moved from mainframe computers to smartphones. Such rapid progress reduces our time to adapt, echoing human developers need to innovate and adapt swiftly. The accelerating pace not only adds weight to the importance of focusing on our irreplaceable human attributes but also amplifies the urgency of preparing for a future dominated by intelligent machines.

The Law of Accelerating Returns not only predicts rapid advancements in AI capabilities, but also suggests a future where AI becomes an integral part of scientific discovery and artistic creation. Imagine an AI agent that could autonomously design new algorithms, test them, and even patent them before a human developer could conceptualize the idea. Or an AI that could write complex music compositions or groundbreaking literature, challenging the very essence of human creativity.

This leap could redefine the human-AI relationship. Humans might transition from being creators to curators, focusing on guiding AI-generated ideas and innovations through an ethical and societal lens. Our role may shift towards ensuring that AI-derived innovations are beneficial and safe, heightening the importance of ethical decision-making and oversight skills.

Yet, theres also the concept of singularity, where AIs abilities surpass human intelligence to an extent where it becomes unfathomable to us. If this occurs, our focus will pivot from leveraging AI as a tool to preparing for an existence where humans are not the most intelligent beings. This phase, while theoretical, imposes urgency on humanity to establish an ethical framework that ensures AIs goals are aligned with ours before they become too advanced to control.

This potential shift in the dynamics of intelligence adds another layer of complexity to the issue. It underlines the necessity for human adaptability and foresight, especially when the timeline for such dramatic changes remains uncertain.

So, we face a paradox: AIs rapid advancement could either become humanitys greatest ally in achieving unimaginable progress or its biggest existential challenge. The key is in how we, as a species, prepare for and navigate this rapidly approaching future.

Featured Image Credit: Provided by the Author; Pexels; Thank you!

I'm an AI engineer and the founder of a pioneering startup in the AI agent development space. My critical approach to analyzing the impact of AI on human developers has been deeply influenced by key works in the field. My reading list spans from Nick Bostrom's "Superintelligence" to "The Age of Em" by Robin Hanson. Through my writings, I aim to explore not just the capabilities of AI, but also the ethical and practical implications it brings to the world of software development.

Original post:
AI Agents: Adapting to the Future of Software Development - ReadWrite

The Race for AGI: Approaches of Big Tech Giants – Fagen wasanni

Big tech companies like OpenAI, Google DeepMind, Meta (formerly Facebook), and Tesla are all on a quest to achieve Artificial General Intelligence (AGI). While their visions for AGI differ in some aspects, they are all determined to build a safer, more beneficial form of AI.

OpenAIs mission statement encapsulates their goal of ensuring that AGI benefits all of humanity. Sam Altman, former CEO of OpenAI, believes that AGI may not have a physical body and that it should contribute to the advancement of scientific knowledge. He sees AI as a tool that amplifies human capabilities and participates in a human feedback loop.

OpenAIs key focus has been on transformer models, such as the GPT series. These models, trained on large datasets, have been instrumental in OpenAIs pursuit of AGI. Their transformer models extend beyond text generation and include text-to-image and voice-to-text models. OpenAI is continually expanding the capabilities of the GPT paradigm, although the exact path to AGI remains uncertain.

Google DeepMind, on the other hand, places its bets on reinforcement learning. Demis Hassabis, CEO of DeepMind, believes that AGI is just a few years away and that maximizing total reward through reinforcement learning can lead to true intelligence. DeepMind has developed models like AlphaFold and AlphaZero, which have showcased the potential of this approach.

Metas Yann LeCun disagrees with the effectiveness of supervised and reinforcement learning for achieving AGI, citing their limitations in reasoning with commonsense knowledge. He champions self-supervised learning, which does not rely on labeled data for training. Meta has dedicated significant research efforts to self-supervised learning and has seen promising results in language understanding models.

Elon Musks Tesla aims to build AGI that can comprehend the universe. Musk believes that a physical form may be essential for AGI, as seen through his investments in robotics. Teslas Optimus robot, powered by a self-driving computer, is a step towards that vision.

Both Google and OpenAI have incorporated multimodality functions into their models, allowing for the processing of textual descriptions associated with images. These companies are also exploring research avenues like causality, which could have a significant impact on achieving AGI.

While the leaders in big tech have different interpretations of AGI and superintelligence, their approaches reflect a shared ambition to develop AGI that benefits humanity. The race for AGI is still ongoing, and the path to its realization remains a combination of innovation, research, and exploration.

Read more:
The Race for AGI: Approaches of Big Tech Giants - Fagen wasanni

Book Review: Re-engineering the Chess Classics by GM Matthew … – Chess.com

Matthew Sadler is a very strong grandmaster (2694 at age 49) and one of the leading computer chess experts. In 2019 he wrote the award-winning Game Changer with Natasha Regan about AlphaZero, and in 2021 he published The Silicon Road to Chess Improvement on how to use chess engines to improve your own game. In addition, Matthew kept the world appraised of the latest engine developments through his tweets and recaps of the Top Engine Chess Championships.

For this latest book Re-engineering the Chess Classics, he teamed up with Steve Giddins to evaluate 40 classical games through the eyes of Stockfish, Leela Chess Zero, and Komodo Dragon. The games are from the period 1852 to 1998 and include games from all the World Champions of that period.

Over the last five years, chess has been revolutionized by the research of AlphaZero, the subsequent implementation of their concepts in Leela Chess Zero, and finally, including neural network technology in Stockfish NNUE. The development of chess engines has been so strong that any opening analysis from before 2020 has lost much of its value. Can the classics stand the test of time?

The themes that emerge from analyzing the forty classics game will not surprise you:

Consider the position after 1.e4 e5 2. Nf3 Nc6 3.Bb5 d6 4.d4 Bd7 5. Nc3 Nge7 6.d5. The engine assessment after 6.d5 is over +2.5 for White, a decisive advantage.

The preference of engines for space has also led to some openings, like the Kings-Indian being hardly playable at the engine level.

Lets assume White moves up his h-pawn. For three tempi (h4-h5-h6), White creates dark square weaknesses on the kingside. The advanced h-pawn restricts the opponents king (mate on g7 but also mate on the back rank). Furthermore, White adds an attacker to his existing attack that might assist other attackers and tie down defenders. Finally, in the endgame, the h7-pawn might become a target.

The advance of the rooks pawn has also impacted the opening theory. For example: 1.d4 Nf6 2.c4 g6 3.Nc3 d5 4.Nf3 Bg7 5.h4 is now a popular Grnfeld Defence variation.

Mistakes come easily in bad positions, but not when you are an engine!

Humans tend to concentrate on one area of the board and devote all their efforts to breaking through on that side, whereas engines are masters at switching plans and creating threats over the whole board.

This was the traditional strength of chess engines and still is.

Interestingly, we play less well than engines because humans play with baggage. In bad positions, we stress out and cannot find the most stubborn defence. When we attack, we focus on breaking through and lack the agility to see the whole board and switch strategy when necessary. Engines play without memory or ego and look with objectivity at every position.

The development of the strongest engines has led to a reevaluation of the relative importance of material, activity, and space. If you want to see how the latest chess concepts impact 40 classics, this book is for you!

The book is currently on introductory offer at ForwardChess for $23.79 and can be pre-ordered at Amazon in hardcover for $34.95.

Continued here:
Book Review: Re-engineering the Chess Classics by GM Matthew ... - Chess.com

The Sparrow Effect: How DeepMind is Rewriting the AI Script – CityLife

The Sparrow Effect: How DeepMind is Rewriting the AI Script

The Sparrow Effect, a term coined to describe the incredible impact of DeepMinds artificial intelligence (AI) technology, is rewriting the AI script and transforming the way we think about machine learning. DeepMind, a London-based AI research lab acquired by Google in 2014, has been at the forefront of AI development, making groundbreaking strides in areas such as natural language processing, computer vision, and reinforcement learning. With its innovative approach to AI research and development, DeepMind is pushing the boundaries of what machines can do and revolutionizing the field of AI.

One of the most notable achievements of DeepMind is the development of AlphaGo, an AI program that stunned the world by defeating the world champion Go player, Lee Sedol, in 2016. Go, an ancient Chinese board game, is considered one of the most complex games in the world, with more possible board configurations than there are atoms in the universe. AlphaGos victory was a watershed moment in AI history, as it demonstrated that machines could not only learn to play complex games but also outperform human experts.

The success of AlphaGo was built on a technique called deep reinforcement learning, which combines deep neural networks with reinforcement learning algorithms. This approach allows AI systems to learn from their own experiences, rather than relying on pre-programmed rules or human input. By playing millions of games against itself, AlphaGo was able to develop its own strategies and refine its gameplay, ultimately surpassing human-level performance.

Following the success of AlphaGo, DeepMind turned its attention to other complex games, such as chess and shogi. In 2017, the company unveiled AlphaZero, an AI system that taught itself to play chess, shogi, and Go from scratch, without any prior knowledge of the games rules or strategies. In a matter of hours, AlphaZero was able to defeat world-class AI opponents in all three games, showcasing the power of deep reinforcement learning and the potential for AI to master a wide range of tasks.

DeepMinds achievements in game-playing AI have far-reaching implications for the broader field of AI research. By demonstrating that machines can learn complex tasks without human intervention, DeepMind has opened the door to a new era of AI development, in which AI systems can learn and adapt to new challenges autonomously. This has the potential to revolutionize industries such as healthcare, finance, and transportation, where AI could be used to optimize processes, make more accurate predictions, and even save lives.

For example, DeepMind has already made significant progress in applying its AI technology to healthcare. In 2018, the company developed an AI system capable of diagnosing eye diseases with the same accuracy as human experts, potentially helping to prevent blindness in millions of people worldwide. Additionally, DeepMind has been working on AI models that can predict the progression of diseases such as Alzheimers and Parkinsons, which could lead to earlier diagnoses and more effective treatments.

Despite the tremendous potential of DeepMinds AI technology, there are also concerns about the ethical implications of AI development. As AI systems become more powerful and autonomous, questions arise about the potential for job displacement, privacy violations, and even the possibility of AI systems making life-or-death decisions. To address these concerns, DeepMind has established an ethics and society research unit, which aims to ensure that AI is developed responsibly and in the best interests of humanity.

In conclusion, the Sparrow Effect, as exemplified by DeepMinds groundbreaking achievements in AI, is rewriting the AI script and opening up new possibilities for machine learning. By pushing the boundaries of what machines can do, DeepMind is not only revolutionizing the field of AI but also paving the way for a future in which AI systems can help us solve some of the worlds most pressing challenges. However, as we continue to explore the potential of AI, it is crucial that we also consider the ethical implications of this powerful technology and work to ensure that it is developed responsibly and for the benefit of all.

Excerpt from:
The Sparrow Effect: How DeepMind is Rewriting the AI Script - CityLife