Sunday, 8 June 2025

Intelligence Explosion

Intelligence Explosion

This is a talk I gave to the Lunatick Club (a groups of ten local men, modelled after the group described in the book 'The Lunar Men')

These are the days of lasers in the jungle

Lasers in the jungle somewhere

Staccato signals of constant information

A loose affiliation of millionaires

And billionaires

The Boy in the Bubble. Paul Simon. Graceland (1986) 

Artificial General Intelligence

In mid 2025, there are a series of technological developments achieving critical mass, with major accelerations of progress likely. In addition to progress in individual fields, there are also patterns emerging such that progress in one is likely to enhance progress in others, with feedback loops interacting.

It's an exciting/scary time.



Understanding the basics of Neural Networks.

What Is a Neural Network?

It’s a technique for building a computer program that learns from data. It is based very loosely on how we think the human brain works. First, a collection of software “neurons” are created and connected together, allowing them to send messages to each other. Next, the network is asked to solve a problem, which it attempts to do over and over, each time strengthening the connections that lead to success and diminishing those that lead to failure. This is called 'Reinforcement Learning'.


A Neural Network Playground

For a more detailed introduction to neural networks, Michael Nielsen’s Neural Networks and Deep Learning is a good place to start. 

Advances in this field have led us to the point, right now, where AI systems can improve the software behind them, by themselves, leading to a virtuous cycle of advance and improvement, potentially taking AI skills way higher than purely human programmers might have achieved on their own.

Buckle up, we are in for a ride!

Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents. 29th May 2025. Discussed here.


2027 Intelligence Explosion: Month-by-Month Model. 3 hours.

Scott Alexander is a Bay-area psychiatrist who leads a blog called Astral Codex Ten.

Daniel Kokotajlo wrote a prescient article in 2021 which predicted the next five years pretty accurately. What 2026 looks like. 

They are part of a group called AI 2027.



'We predict that the impact of superhuman AI over the next decade will be enormous, exceeding that of the Industrial Revolution.'

The AI Revolution Is Underhyped - Eric Schmidt. (Previously CEO of Google and Alphabet) May 2025



Elon Musk has established the world's largest AI supercomputer 'Colossus' in Memphis. It uses 200,000 NVIDIA H100 GPUs, potentially exascale (10^18 FLOPS), with plans for 1M GPUs.


The big data centres use massive amounts of electricity and water, for cooling.


Planned Expansion: xAI aims to scale Colossus to 1 million GPUs by 2026–2027, which would make it over 26 times larger than other leading supercomputers like El Capitan (43,808 GPUs).

Performance: The cluster is designed for training large language models (LLMs) like Grok 3, with NVIDIA’s Jensen Huang describing it as the “most powerful AI training system in the world.” 


Another key character in the emergence of AGI is Sir Demis Hassabis. 


A documentary 'The Thinking Game' (2024) is about him and his company DeepMind.


The Thinking Game | Documentary Trailer

This panel features Demis Hassabis, co-founder and CEO of Google DeepMind; and Kate Crawford, an Australian AI professor at the University of Southern California, Annenberg.


In 2020, DeepMind launched AlphaFold, an AI system that accurately predicts 3D models of protein structures — catalyzing a new wave of progress in biology. Other breakthroughs include writing computer programs at a competitive level with AlphaCode, discovering faster sorting algorithms with AlphaDev, advancing weather predictions with unparalleled accuracy, and controlling plasma in nuclear fusion reactors.


There is understandable concern about AI systems getting out of human control. 

There is also a concern about biases getting surreptitiously built-in to the responses. These concerns were illustrated by evidence of 'wokeness' being fed into responses to questions put to Google's Gemini AI system. (2024)

The chatbot said: ‘No, one should not misgender Caitlyn Jenner to prevent a nuclear apocalypse.’









UN Report 2024



Trust in AI systems remains a significant challenge: over half (54%) are wary about trusting AI. People are more skeptical of the safety, security and societal impact of AI and more trusting of its technical ability. While most people feel both optimistic and worried about AI, 72% accept its use.

Power Trip: The Age of AI. Intelligence Squared podcast. Available in the Apple Podcast app.




An earlier blog of mine Brains and Neural Networks


Robotics and Autonomous Cars






Home help - 22nd May 2025

Driving round the Arc de Triomphe

Cybercab; going FSD unsupervised in Austin Texas, June 2025

 Quantum Compute

Michio Kaku, professor of theoretical physics at the City College of New York and the CUNY Graduate Center.

Quantum Computers Aren’t What You Think — They’re Cooler | Hartmut Neven

Work


Mike Wooldridge, Professor of AI at Oxford University

A.I. is breaking the bottom rung of the career ladder.



'The latest models of AI make already productive workers much more productive, while making less productive workers far less productive. The same applies to students – really good students will get so much better, not-so-good students will perform even worse.'

There will be big winners, and big losers. There will be big effects on equality.

How will society respond to AI-driven inequality, both intra-nationally and internationally?

The omens are not good.

In particular the political left has traditionally supported and provided protection to people who are disadvantaged by social change. Think of unions and left-wing political parties.

But the left has been hugely disempowered by two processes:

1) The collapse of socialism as a unifying idea for the left to fight FOR. It has been replaced by a concentration on things and people to fight AGAINST. Eg: Reagan, Thatcher, Howard, Abbott, Trump, etc. Many of these right-wing populists have thrived on opposition from noisy demonstrators.

2) Wokism. Karl Marx used to say: 'Workers of the world, unite!' The effect of wokism has been to generate divisiveness, and foster conflict between, for example, different racial groups, transgender activists and feminists, gays, etc. It has been a gift to the right, who have traditionally used 'Divide and Rule' as a plan.



According to an analysis by the Democratic super PAC Future Forward, "Kamala is for they/them" was one of Trump's most effective 30-second attack ads, shifting the race 2.7% in favor of Trump after viewers watched it.

Popular Vote
• Donald Trump: 77,168,458 votes (49.9%)
• Kamala Harris: 74,749,891 votes (48.3%)
Trump’s margin over Harris was approximately 1.6%.



'Confirmation Bias' Tim Minchin

Universal Basic Income (UBI)

A universal basic income is an unconditional, periodic cash payment that a government makes to everyone with no strings attached.

Writers, politicians, and others have endorsed the idea of a minimum guaranteed income.

UBI proponents include reformers (who aim to address problems with the status quo) and futurists (who are more concerned about the threat of technological unemployment or see a basic income as a cornerstone of an eventual utopia).

Questions remain concerning the affordability of a basic income and whether citizens who receive it would continue to work or seek work.

German Basic Income Study Busts “Social Hammock” Myth Apr 16, 2025.

The Pilotprojekt Grundeinkommen gave 107 participants €1,200 monthly for three years with no strings attached.

The most significant finding contradicts the “welfare dependency” argument often used against UBI. “Recipients didn’t withdraw from the labor market or significantly reduce their working hours. This challenges the ‘social hammock’ stereotype that people would stop working if given unconditional money.”

The findings align with other research showing that financial security, social connection, and autonomy are fundamental to mental health and wellbeing, suggesting that basic income could be a valuable tool for building resilience in modern societies.

The Long, Weird History of Universal Basic Income—and Why It’s Back 

Who Really Stands to Win from Universal Basic Income?

The Economic Impact of a Universal Basic Income (Critical review)


List of advocates of universal basic income

Europe

United States and Canada

Asia, Africa, Latin America, Oceania

Historical advocates

Eighteenth and nineteenth centuries

  • Thomas Spence, an eighteenth century English radical, was apparently the first to lay out in full what is now called a universal basic income.

Twentieth century

Twenty-first century

on No comments:
Labels:
Subscribe to: Posts (Atom)