in pages Frontiers in human neuroscience He published a post dedicated to the reaction of the human eye to optical illusions. However, it seems that they can also play an important role in training AI.
A few weeks ago, we reported an optical illusion that gives us the impression that a black hole is absorbing us. In fact, the image on a piece of paper or a screen, of course, remains constant, but our brain interprets it differently.
The brain has no way of knowing what [naprawdę] exist. What it does is build a kind of virtual reality around what might be out there. There is a little guesswork in this. In this respect, you can think of the brain as a kind of probabilistic machine. […] He uses some previous hypotheses and tries to test them all the time to see if they work.Bruno Ling from the University of Oslo explains
As the lead author of the study adds, by constricting the pupil, the human body adapts to a very likely short-term situation. For example, sunlight can temporarily incapacitate us. This will have serious consequences, whether we are a victim or a predator in a given situation. Missing a precious millisecond can be fatal.
Optical illusions determine how advanced a system is
We also need to predict the future, even if it is not for amazing uses, such as, for example, guessing the numbers that will be drawn in the lottery. It is about predicting how a particular object, such as a volleyball, will behave. If we were to follow her flight path, without guessing where she was likely to be in a moment, we wouldn’t have much chance of bouncing back or catching up with her.
Therefore, the brain anticipates what might happen. And if a robot is to function effectively in the real world, it must also anticipate developments and make adjustments on an ongoing basis. It should be very easy for computers to be able to perform calculations at an exponential rate, much faster than the human brain. However, man has millions of years of evolution behind him.
Thus, the road for machines to reach the level at which people interpret the environment will be long, although it will take a little less than a million years. Optical illusions may be the basis of success. Thanks to them, scientists will be able to determine how well machine vision systems perform compared to biological organs. If such a system falls into imaginary traps similar to the human brain, that would be an indication of its complexity.
“Infuriatingly humble musicaholic. Problem solver. Reader. Hardcore writer. Alcohol evangelist.”