To trust or not to trust the AI.
I do not fear computers. I fear the lack of them. Isaac Asimov
I live in Buenos Aires. Every day I drive to work, it’s really not a long journey, however in the last decade the city got really full of cars and any accident or problem could easily double the time of my trip. I usually go by the same route but I always got a couple of alternative routes in mind just in case.
I started using Waze to get driving directions as it estimates the travel time, shows traffic jams and obstacles reported by the community. I think it's a wonderful app, I’m quite a fan of it although I realized that it took me some time to stop thinking and trusting my decisions on this little piece of technology.
At first, I found myself many times overriding the recommended route by the app and going my own way believing I knew the city better. Once, I crashed after taking a detour from the suggested route ☺(hopefully no harm done!). After a certain time, I started to trust in the recommendations as it consistently got me faster to destination and now I practically blindly follow its suggestions.
These personal experiences made me think about our relationship with technology and specifically with expert systems and artificial intelligence. Apps like Waze that assist you on an specific real-life use case will start to appear on different aspects of our decision making.
Let’s see how we got to this.
The Genesis
The genesis of the Internet consisted on the interconnection of a few computers owned by universities and other organizations for research purposes.
Some years later it took a big leap forward with the creation and popularization of the personal computer as it allowed the network access of millions of devices and with a much more general purpose, closer to the average consumer needs.
Then the Internet developed senses, mobile devices: digital cameras, cellphones, GPS, weather sensors, wearables, etc. collecting information about our everyday life and plugging it to the network. We are the agents that connect the real world to the Internet through these devices, feeding it with a staggering 5 exabytes of data every day from our interaction with the services we use.
So what will the future look like? The number of sensors will keep growing faster and faster as your home appliances, pets, cars, watches, homes, lights, connects to the network in a concept called the "Internet of Things". According to Cisco, there will be around 50 billion "things" connected by 2020.
Getting Smart
Today the Internet is a powerful tool as it let us instantly get information from all around the world and share our voice with the same ease, however, data in most of the cases is still used by humans to take our own decisions.
Computer science is in the search for a generalized pattern recognizer (see Deeplearning) that could be applied to many different scenarios to create Artificial Intelligence and set the base for the next stage in the evolution and help create a Smart Network. When I say smart I mean that it will gradually take more and more decisions for us in different aspects of our lives.
In the future, the AI will take into account many more factors when making predictions and identifying patterns that we humans cannot see, enabling it to make better judgments. As a consequence we will eventually trust the AI to offload a good part of our own decisions.
Some questions I have regarding the future coexistence between humans and AI. Is it possible for an AI to develop cognitive biases as well as we human do? Will there be one general super-intelligent AI or we will have multiple AI with different "personalities" existing on the Internet? Will we educate AIs and then they’ll educate new generations of humans? How we will handle privacy issues, when simultaneously, sharing is key for AIs to help us making better decisions?
I’ll write more thoughts about this on the following posts ☺
