Will AI Save Mankind?

Only if mankind decides to save itself will it be saved.  So far, with our satanic regime trying to kill us all, the odds of survival are plunging without protest from our captured media. And if the warmongering should abate for some reason, economic self-destruction is guaranteed because of our ever-counterfeiting central bank and the ever-growing mother lode of government debt.

Meanwhile, almost quietly, AI is exponentially working its way into our lives as a useful tool.  Will it ever do more?  Could one of them pass the Turing Test today?  If so, could it go further and bring an end to wars and economic self-destruction?  The path of intelligence as laid out by human experts is Narrow AI, General AI, and Super AI.  But it seems to be flying past the General stage.  Having AI roughly equal to us was thought to be a perceptible stage of development.

But maybe not. How to Fight Artificia... Cutler, Boone Best Price: $23.40 Buy New $30.50 (as of 03:21 UTC - Details)

In March 2016, over a period of six days, AlphaGo defeated Lee Se-dol, a professional Go player of 9 dan rank (highest ranking), 4-1, in a match broadcast live throughout South Korea. AlphaGo, an AI developed by Google’s DeepMind, proved conclusively that underlying neural networks could solve problems of incomparable complexity.

The significance of the victory was such that Thore Graepel, one of the scientists who built AlphaGo, declared “the “Holy Grail” of AI had been found.”

“People often say there are more positions in the game of Go than there are atoms in the known universe,” he said.

“But the truth is that if for every atom in the known universe, you had another universe and [if] you counted all the atoms in that collection of universes, that comes closer to the number of positions in the game of Go.”

Brute-force computation, the kind embedded in IBM’s Deep Blue that defeated chess grandmaster Gary Kasparov in 1997, would not work playing Go.  It’s not possible to consider every possible move.  According to DeepMind, what was needed was a computer that could “mimic the human quality of intuition.”  With Go’s nearly infinite possible moves, the AI somehow had to know which moves to consider and which to discard.

For this, it used a technique known as machine learning, which is much like human learning.  According to AI pioneer Arthur Samuel, machine learning is “the field of study that gives computers the ability to learn without explicitly being programmed.”  Instead of giving a computer specific instructions for solving a given task, such as feeding it descriptions of a cat or dog, in machine learning computers are left to figure things out on their own, through observation.  Show a machine-learning computer enough pictures of cats and dogs and it will recognize new pictures of cats and dogs without being given a definition.

When Lee Se-dol sat down to play AlphaGo in a five-star hotel in Seoul he was confident the AI was out of its league.  Three-and-a-half hours later AlphaGo won, and Mr. Se-dol was left staring at the board in stunned disbelief.

Game two the following day became a turning point in AI history.  On the 37th move, AlphaGo made a move so unlikely “no human in a million years would have thought of it,” said Chris Carlock, a competent Go player and the English commentator during the match.  But move 37 turned out to be ingenious.  It didn’t learn the move from watching humans play.  It thought it up on its own.  It had invented a winning move.

“AlphaGo’s Move 37 has become a symbol of machine creativity, commemorated on mugs and t-shirts.”  Sadly, neither are currently available on Amazon.

In game four, the only game Mr. Se-dol won, he returned the favor playing an unlikely move that confused AlphaGo.  It was the last time a human has defeated DeepMind’s stellar AI.

Demis Hassabis, chief executive and co-founder of DeepMind, congratulated “Lee Se-dol for his legendary decade at the top of the game, and wished him the very best for the future.”  In 2019 Mr. Se-dol retired because AI, he said, “cannot be defeated.”

A year after its match with Mr. Se-dol, AlphaGo Zero knocked AlphaGo from its pedestal, defeating it 100 games to 0.  As DeepMind explains:

Previous versions of AlphaGo initially trained on thousands of human amateur and professional games to learn how to play Go. AlphaGo Zero skips this step and learns to play simply by playing games against itself, starting from completely random play.

To accomplish this it used a novel form of reinforcement learning, in which Zero became its own teacher.

This technique is more powerful than previous versions of AlphaGo because it is no longer constrained by the limits of human knowledge. Instead, it is able to learn tabula rasa from the strongest player in the world: AlphaGo itself.

After three hours of training Zero became a competent amateur. After 70 hours of training, playing many millions of AlphaGo vs. AlphaGo games, it achieved a superhuman level of play in Go.

DeepMind tackles protein folding problem

Proteins, found inside every cell of the body, are considered the building blocks of life.  But figuring out the exact structure of a protein can sometimes take years, meaning scientists are only able to study a few of them.

Following its Go victory in 2016, DeepMind spent four years developing AlphaFold “by showing it the sequences and structures of around 100,000 known proteins.”  By 2020, AlphaFold 2 was ready for CASP, “a biennial challenge for research groups to test the accuracy of their protein structure predictions against real experimental data.” The Age of AI: And Our... Huttenlocher, Daniel Best Price: $6.55 Buy New $12.51 (as of 06:59 UTC - Details)

AlphaFold 2 “demonstrated a level of accuracy so high that the community considered the protein–folding problem solved.”

In 2024, along with Isomorphic Labs, DeepMind released AlphaFold 3, “which predicts the structure and interactions of all of life’s molecules.”

Conclusion

Ada Lovelace is considered the world’s first computer programmer for her work on Charles Babbage’s Analytical Engine.  Alan Turing quoted her in his 1950 seminal paper, Computing Machinery and Intelligence: “The Analytical Engine has no pretensions to originate anything. It can do whatever we know how to order it to perform” (her italics).”  According to the evidence available to Lady Lovelace, Turing says, he would agree.

AI will be “more profound than fire or electricity” said Alphabet’s CEO. And like those forces it’s imperative to create AI with a generous serving of caution and public understanding if we are to direct it toward our earthly salvation.