How much time does it take to play a video game?
Really? There isn't a punchline for that. Thanks to a neural network system, there is an actual answer.
The number of brain cells in that game is close to one million.
Their one-sided strategy for digital table tennis won't see them win any e-sports titles in the near future, but it does show the potential in using Silicon technology.
This is the first synthetic biological intelligence experiment that shows that neurons can learn to perform a task better if they are given feedback. It's amazing, with potential applications in computing, as well as studying all sorts of brain stuff, from how drugs and medication affect brain activity to how intelligence develops in the first place.
"We have shown that we can interact with living biological neurons in a way that makes them modify their activity, which leads to something that resembles intelligence," says neuroscience researcher and founder of Cortical Labs.
Stem cells and embryonic cells are used in the creation of DishBrain. The cells were grown on an array of microelectrodes that could be stimulated to the point of providing sensory input.
If the ball was to the left or right of the paddle, signals from the frequencies on either side of the dish would tell you.
The set-up allows DishBrain to move the paddle to meet the ball, but it doesn't work out very well. The game needs feedback to be played well.
Whenever DishBrain missed the ball, the team developed a software to give feedback. The system was able to improve at playing pong with the help of this.
Karl Friston is a theoretical neuroscientist at University College London.
The cultures were able to make their world more predictable by acting on it. You can't teach this kind of self-organization because these mini brains don't have a sense of reward or punishment.
Friston developed a theory called the free energy principle, which proposes that all biological systems behave in ways that reduce the gap between what is expected and what is experienced to make the world more predictable.
Friston says that DishBrain is doing what biology does best by adjusting its actions.
One of the first games used in machine learning was Pong.
The system as a whole would reorganize its activity to better play the game and minimize having a random response when an unpredictable stimuli was applied to the cells. Playing the game, hitting the ball and getting predictable stimulation is creating more predictable environments.
There are some really intriguing possibilities in this. Our best computers can't duplicate the power of the human brain, which contains up to 100 billion cells. Just one second of the activity of one percent of the human brain was recreated with the help of 82,944 processors and a petabyte of main memory.
This goal may not be out of reach if the architecture is similar to an actual brain.
There are more immediate ramifications.
It is possible that DishBrain can help chemists understand the effects of various medications on the brain. One day, it might be possible to reverse engineer stem cells from a patient's skin to create a different type of brain cell.
Friston says that it means we don't have to worry about creating digital twins to test therapeutic interventions. "We now have the ultimate biomimetic'sandbox' in which to test the effects of drugs and genetic variant on your brain and my brain."
The next step is to figure out how drugs and alcohol affect a person's ability to play a board game. "We're trying to create a dose response curve with the idea of getting them drunk and seeing if they play the game worse than before."
The brain cells roll into a bar.
The research was published in a journal.
All Rights Reserved © 2024
