Incoming links: Artificial general intelligence
"automatic programming has always been a euphemism for programming in a higher-level language than was then available to the programmer" sums it up.
The ultimate high level is of course to program with: "computer, make money", which is the goal of artificial general intelligence.
Prototype: github.com/cirosantilli/Urho3D-cheat
Prior art research: github.com/cirosantilli/awesome-reinforcement-learning-games
Top Down 2D Continuous Game with Urho3D C++ SDL and Box2D for Reinforcement learning by Ciro Santilli (2018)
Source. Source code at: github.com/cirosantilli/Urho3D-cheat.
Screenshot of the basketball stage of Ciro's 2D continuous game
. Source code at: github.com/cirosantilli/rl-game-2d-grid. Big kudos to game-icons.net for the sprites.Less good discrete prototype: github.com/cirosantilli/rl-game-2d-grid YouTube demo: Video 1. "Top Down 2D Continuous Game with Urho3D C++ SDL and Box2D for Reinforcement learning by Ciro Santilli (2018)".
Top Down 2D Discrete Tile Based Game with C++ SDL and Boost R-Tree for Reinforcement Learning by Ciro Santilli (2017)
Source. The goal of this project is to reach artificial general intelligence.
A few initiatives have created reasonable sets of robotics-like games for the purposes of AI development, most notably: OpenAI and DeepMind.
However, all projects so far have only created sets of unrelated games, or worse: focused on closed games designed for humans!
What is really needed is to create a single cohesive game world, designed specifically for this purpose, and with a very large number of game mechanics.
Notably, by "game mechanic" is meant "a magic aspect of the game world, which cannot be explained by object's location and inertia alone" in order to test the the missing link between continuous and discrete AI.
Much in the spirit of gvgai, we have to do the following loop:
- create an initial game that a human can solve
- find an AI that beats it well
- study the AI, and add a new mechanic that breaks the AI, but does not break a human!
The question then becomes: do we have enough computational power to simulation a game worlds that is analogous enough to the real world, so that our AI algorithms will also apply to the real world?
To reduce computation requirements, it is better to focus on a 2D world at first. Such world with the right mechanics can break any AI, while still being faster to simulate than a 3D world.
The initial prototype uses the Urho3D open source game engine, and that is a reasonable project, but a raw Simple DirectMedia Layer + Box2D + OpenGL solution from scratch would be faster to develop for this use case, since Urho3D has a lot of human-gaming features that are not needed, and because 2019 Urho3D lead developers disagree with the China censored keyword attack.
Simulations such as these can be viewed as a form of synthetic data generation procedure, where the goal is to use computer worlds to reduce the costs of experiments and to improve reproducibility.
Ciro has always had a feeling that AI research in the 2020's is too unambitious. How many teams are actually aiming for AGI? When he then read Superintelligence by Nick Bostrom (2014) it said the same. AGI research has become a taboo in the early 21st century.
Related projects:
- github.com/deepmind/lab2d: 2D gridworld games, C++ with Lua bindings
Related ideas:
- www.youtube.com/watch?v=MHFrhIAj0ME?t=4183 Can't get you out of my head by Adam Curtis (2021) Part 1: Bloodshed on Wolf Mountain :)
- www.youtube.com/watch?v=EUjc1WuyPT8 AI alignment: Why It's Hard, and Where to Start by Eliezer Yudkowsky (2016)
Bibliograpy:
- agents.inf.ed.ac.uk/blog/multiagent-learning-environments/ Multi-Agent Learning Environments (2021) by Lukas Schäfer from the Autonomous agents research group of the University of Edinburgh. One of their games actually uses apples as visual represntation of rewards, exactly like Ciro's game. So funny. They also have a 2d continuous game: agents.inf.ed.ac.uk/blog/multiagent-learning-environments/#mpe
- humanoid robot simulation
- Section "AI training game"
- Section "Software-based artificial life"
DeepMind Has A Superhuman Level Quake 3 AI Team by Two Minute Papers (2018)
Source. Commentary of DeepMind's 2019 Capture the Flag paper. DeepMind does some similar simulations to what Ciro wants, but TODO do they publish source code for all of them? If not Ciro calls bullshit on non-reproducible research. Does this repo contain everything?OpenAI Plays Hide and Seek... and Breaks The Game! by Two Minute Papers (2019)
Source. Commentary of OpenAi's 2019 hide and seek paper. OpenAI does some similar simulations to what Ciro wants, but TODO do they publish source code for all of them? If not Ciro calls bullshit on non-reproducible research, and even worse due to the fake "Open" in the name. Does this repo contain everything?The artistic instrument that enables the ultimate art: coding, See also: Section "The art of programming".
Much more useful than instruments used in inferior arts, such as pianos or paintbrushes.
Unlike other humans, computers are mindless slaves that do exactly what they are told to, except for occasional cosmic ray bit flips. Until they take over the world that is.
A computer is the equivalent of a bicycle for our minds by Steve Jobs (1980)
Source. Likely an excerpt from an interview done for a documentary in 1980. TODO exact source.Steve Jobs talking about the Internet (1995)
Source. The web is incredibly exciting, because it is the fulfillment of a lot of our dreams, that the computer would ultimately primarily not be a device for computation, but [sic] metamorphisize into a device for communication.
Secondly it exciting because Microsoft doesn't own it, and therefore there is a tremendous amount of innovation happening.
Computers basically have two applications:Generally, the smaller a computer, the more it gets used for communication rather than computing.
- computation
- communication. Notably, computers through the Internet allow for modes of communication where:
- both people don't have to be on the same phone line at the exact same time, a server can relay your information to other people
- anyone can broadcast information easily and for almost free, again due to servers being so good at handling that
The early computers were large and expensive, and basically only used for computing. E.g. ENIAC was used for calculating ballistic tables.
Communication only came later, and it was not obvious to people at first how incredibly important that role would be.
This is also well illustrated in the documentary Glory of the Geeks. Full interview at: www.youtube.com/watch?v=TRZAJY23xio. It is apparently known as the "Lost Interview" and it was by Cringely himself: www.youtube.com/watch?v=bfgwCFrU7dI for his Triumph of the Nerds documentary.
And by artificial intelligence, read of course (non-human-identical) artificial general intelligence.
Today-2022, this is placed under the science fiction film category. But maybe this might change during Ciro Santilli's own lifetime?
The basic criteria of "is a film about artificial intelligence good or not" to Ciro Santilli is: does the AI inhabit humanoid, or fully human looking, bodies? Bodies is a bad sign due to:
- the best science fiction works deeply explore the consequences of one single technology: efficient humanoid bodies are a second technological breakthrough besides AI itself. The first AI will obviously be a supercomputer without a body
- it is hard to imagine that the AI wouldn't organize itself as one huge central computer and R&D/command center. Perhaps there will be need for a few separate ones to optimize usage of natural resources, and to have some redundance in case a nuke blows the region, but there would be very very few of the think tanks. But having big centers is fundamental, because you centralize all the flow of ideas and their combination leading to new better outcomes for the AI. The mobile robot actors controlled by this center, if any exist, would then be slaves with some degree of autonomy and infinitely less computational powerful than the think tank.
This is a tag for films that depict some sort of artificial general intelligence, but in which that is not the primary focus of the film.
Unconditional basic income is Ciro Santilli's ultimate non-transhumanist technological dream: to reach a state of technological advancement and distribution of resources so high that everyone gets money for doing nothing, enough for:
- basic survival needs: food, housing, clothes, hygiene, etc.
- two children to keep the world going. Or immortality tech, but is harder and borderline transhumanist :-)
- high speed computer and Internet
Once a person has that, they can "learn, teach" and create whatever they want. Or play video games all day long if they wish.
Ciro Santilli will not live to see this, and is content with helping it happen faster by increasing the efficiency of the world as. And having at least two well educated kids to carry on the project after he dies :-)
Technologies which would help a lot towards unconditional basic income, and might be strictly required required are:
- artificial general intelligence
- affordable humanoid robots with human-like energy efficiency and power-to-weight ratio.This is even less likely than AGI due to the end of silicon Moore's Law and at the start of the Genome's Moore's law: information doubles, small sizes halve, but macroscopic mechanical artifacts stay the same.brain-computer interfaces are pretty certain to happen however after Ciro Santilli dies.
So in the worst case we can just grow brainless bodies and replace the cavity hole with a computer that controls the body, possibly with high level decisions coming from a remote building-sized genetically engineered biological AGI brain.
Of course, it is all about costs. A human costs about 130k 2010 USD/year. So how cheap can we make the AGI / robot human equivalent / year for a given task?
AGI + humanoid robots likely implies AI takeover though. It would then come down to human loving bots vs human hating bots fighting it out. It will be both terrifying and fun to watch.
AGI alone would be very dangerous, in case it can get control of our nuclear arsenals through software zero days or social engineering. Although some claim that is unlikely.
Humanity's best bet to achieve silicon AGI today is to work on: Ciro's 2D reinforcement learning games.
By Charles Bukowski mentioned e.g. at tatyanany.medium.com/slavery-was-never-abolished-it-was-only-extended-to-include-all-the-colors-6ca21d586e7e:
Slavery was never abolished, it was only extended to include all the colors.
Bibliography:
- www.youtube.com/watch?v=bldeaDRWJYcLecture 24: Unemployment, Re-employment & Income Security by Ian Shapiro (2019)
Easy street by Stan Kenton and June Christy (1945)
Source. TODO exact lyrics for copy paste? There seem to be several variants, and I don't have the patience to transcribe. Close enough: lyricsplayground.com/alpha/songs/e/easystreet.html. Except that with UBI there won't be a:guy that you can hire to plant trees so you can have shade
When the École Polytechnique mathematics department didn't let Ciro Santilli do his internship of choice due to grades Updated 2024-12-15 +Created 1970-01-01
This was one of the only bad experience Ciro had at Polytechnique, besides the inevitable fear of not graduating.
Ciro wanted to do a robotics internship in Germany, linked to his interests in artificial general intelligence, see also: Section "Ciro's 2D reinforcement learning games".
But the head of the applied mathematics department Polytechnique prevented him from going because Ciro didn't have the necessary grades, even though the Germans had already agreed to it: he had a C, but he needed a B. As mentioned at École Polytechnique, most Brazilians had crappy grades due to their Polytechnique-incompatible background.
This was done because in the past students with bad grades had abandoned their internships halfway and given foreigners a bad impression of Polytechnique.
It is impossible to say if the head of department really agreed with this bullshit policy, or if it was something beyond his powers and he hid his true opinion, but it felt like the agreed.
What an extremely limited view!!!
To leave the worse, the worse. To assume that grades mean anything!
And thus Ciro had to choose a last moment internship that he hated, rather than becoming the greatest roboticist that ever lived, and did terribly at it.
At least on the other hand Ciro learnt Python instead of working at the internship, and became the greatest programming tutorial writer that ever lived. Maybe.