Not all the work in AI is for the TV to set the exact channel according to who requests it.
The alarm clock sounds 6 27 am, I tell it I turned it off and it “knows” that in 10 minutes it should ring again, I turn it off, I greet my cell phone and open the browser to make the morning news available, I’m going to take a shower, the coffee machine turns on 6 55 am, I finish taking a shower and I drink the coffee while I dress, I walk to the door and when I take the keys the list of reminders for the day is reproduced.
You can continue to narrate the day of any average person and the relationships with the “machines” in his daily life and, more or less accurate in each case, you will see the level of use of technology in which we are immersed and we have naturalized. The one that gives us comfort, saves us time and problems, and with which we collaborate so that it can continue to improve.
What’s the next step?
A few days ago I read a note that talked about the basic skills where Google is based to compete with the other two big ones (Apple and Microsoft), these are machine learning, deep learning and AI. The interesting thing is not in itself “what”, but “how” you are working from the learning (machine learning and deep learning) and the user experience for improvement in AI.
TensorFlow, the machine learning system with which you are working, is built on the basis and error correction of your DistBelief precursor and maintains the same philosophy with which you expect the user to collaborate. To take the experience to develop artificial intelligence technology by minimizing the error.
The challenge of all this is not how the coffee taste and pleases the person who gave the order, but that a car can conduct itself, relate to everyone else that circulates around the city and improve mobilization in the city.
Meanwhile we continue to collaborate with our feedback in the daily use of the machines.