The climatic conditions do the same in a desert city where the sun always shines on average there are 17 rainy days a year the sensors mount on the robotaxis do not run the risk of having visibility problems due to the rain of being confus by leaves blown by the wind or going haywire in fog. How would an autonomous car abandon in the confusing alleys of the center of Milan react in the midst of scooters whizzing by in Rome in the rain in London or in the traffic of Istanbul For the moment no startup has even dar to test its vehicles in the. Until a few years ago however it seem it was only a matter of time after the first experiments in the metropolis with wide lanes and good climatic conditions they would.

Gradually acquire the skills necessary to

Drive even in the most chaotic environments. With the passing of time however we are realizing that it is not possible to single out the best way to Italy Phone Number List deal with each situation that occurs in city traffic in the real world purely through statistics where the unknowns unexpect events unexpect events and infractions are the order of the day. As Volvos head of autonomous driving Marcus Rothoff recently admitt the randomness of behavior cannot be handl by todays technology.

Back to index Deep learning more quality

There are other aspects to be evaluat and which make us fear that the potential of deep learning is being push to its limit hypothesis support B2C Phone List by the aforemention Gary Marcus but not for example by Yann LeCun winner of the. Turing Award for his role in the invention of deep learning itself. For example to obtain significant progress it is now necessary to increase the computational capabilities of the system much higher than the improvements obtain. In short the returns are decreasing and the costs consequently grow more and more to the point that it is estimat that the training of the OpenAI GPT3 model alone cost over 12 million dollars.