Humans and machines combining to produce superhuman intelligence can soon be a thing of science fiction, according to the US Space Force’s chief scientist, who predicts it will happen in the “coming decade.”
Dr. Joel Mozer declared at an Air Force Research Laboratory event on Wednesday that we are approaching the era of ‘human augmentation,’ which is critical to the US national security in order to avoid falling behind our strategic competitors.
His idea, though, would not include the transformation of humans into cyborgs, but rather the use of “AI agents” to aid in strategic military preparation.
Mozer points to the skills shown by AlphaGo Zero, a Google affiliate that was able to teach itself to learn the game of Go in just a few weeks.
Extortionate capabilities, according to Mozer, can contribute to superhuman capabilities by mixing human imagination with computer strength, speed, and performance.
The Pentagon would have access to a broader range of ideas than any human or humans could fathom on their own.
The Pentagon will have access to a wider variety of concepts than any individual person or group of humans might imagine.
Humans “resolve uncertainty, apply judgment, philosophy, and ethics,” while AI “crunch the possibilities.”
“The results would be superhuman,” explained the Space Force physicist.
AlphaGo Zero is an improved variant of AlphaGo, but it is self-taught, unlike its predecessor.
Only the rules of chess were programmed into the AI, and it played games against itself using reinforcement learning before it was able to predict its own moves.
“After thousands and thousands of moves, understood how those moves could impact the game,” Mozer said.
After more than three million games, AlphaGo Zero beat its predecessor, AlphaGo, who had benefited by studying the skills of the world’s best chess players.
AlphaGo Zero “was completely unbeatable” after 21 days.
“It developed new and innovated strategies. Some of which we don’t understand,” stated Mozer.
“That is a little bit scary. Sometimes machine learning algorithms reach their objectives in ways that we don’t necessarily comprehend, but imagine if we using these technologies, capabilities, wisely, in our military systems.”
The Pentagon could “develop strategies and tactics that no human could,” implying that AI could one day issue orders to top generals, if those inventions are used in strategic military planning.
The technology would devise theories we’d never seen before and present them to people, resulting in superhuman intellect.
The military would have access to a wider variety of ideas than any individual person or group of humans could imagine.
“This will extend to the battlefield, where commanders and decision makers will have at their disposal multiple autonomous agents, each able to control the execution of things like reconnaissance, or fire control, or attack,” Mozar said while warning that “we must think carefully about the ethics of this, and how we will trust these autonomous agents, especially in an era of lethal autonomous warfare.”
Although there are risks associated with using ‘AI agents,’ Mozar clarified that US adversaries are still operating on similar inventions, and ‘”we cannot afford to lag.”
The invention could also create a ‘”superhuman workforce,'” in addition to superhuman intellect on the battlefield.
“Augmented reality, virtual reality, and nerve stimulation” will be used by this team.
“You could put [an] individual into a state of flow, where learning is optimized and retention is maximized,” Mozar said. “This individual could be shaped into somebody with very high-performing potential.”
The Space Force scientists envision humans and machines working together as a team, but the US Army is working to merge the two.
Experts from Devcom (the Combat Capabilities Development Command) presented a range of potential developments that could be used to improve soldiers on the battlefield by 2050 in 2019.
Enhanced limbs with greater mobility, an eye with infrared and ultraviolet vision, and an auditory system with ultra- and subsonic hearing are among them.
They also propose that a potential soldier could be equipped with a neural implant that boosts brain strength and allows them to manipulate weapons with their minds.
Hundreds of physicists, military officers, ethicists, and other scholars participated in the “thought experiment,” debating future technology, the effect of cyborgs on humanity, and how they might affect warfare.