With a new year starting, you would have already seen a lot of industry experts sharing their insights on the future of AI and considering 2020 to be the year when artificial intelligence will become a reality. However, the reality is quite the opposite of what non-tech people imagine.
AI machines that we have created today are nothing but a machine automating some processes. The word artificial intelligence isn’t fit for these types of systems. But except all the buzz cycle about robotic dogs, chatbots, virtual assistants, and automated machines there is nothing that could make these systems intelligent.
The companies, on the other hand, see it from a different perspective. For them, data is what matters the most. Collecting data and using it to generate more leads, is what they are interested in.
Making AI truly intelligent
To make artificial intelligence intelligent, we need to make them think like humans and process information like our brain does. For this purpose, a large amount of data isn’t important, what it requires is correct data and processing neural networks that could replicate human minds.
One of the things that make the human mind work intelligently is its ability to forget and learning. It processes information in real-time and replaces it with the previous information. For instance, our mind could easily identify a cat by seeing it once, unlike AI, which processes thousands of images of animals to recognize the cat.
This ability of our mind help us making decisions in real-time even without experiencing the same situation ever before.
Artificial intelligence that constantly reevaluates the previously learned knowledge, as the human mind would, will be the true intelligence. If this quality to forget things and reevaluating past information is implied to AI, it will be able to identify the most relevant data and make the decision in real-time with a low computational cost.
However, this quality isn’t enough alone. Combining different human abilities will make true Artificial General Intelligence (AGI) possible.
Fusing the strengths of Neural and Symbolic AI
The deep neural networks are great for collecting and processing a vast amount of data, however, they lack abstraction. On the contrary, symbolic AI is good at abstraction but couldn’t process large streams of data.
By combining, strengths of both deep nets and symbolic AI, advancement could be made towards the goal of achieving AGI.
Interestingly, the most successful innovations in the field of AI are through the fusion of these technologies. Google’s AlphaGo is a good demonstration of this combination. Furthering the cause, Google and other tech companies are focusing on developing true artificial intelligence, ensuring that their investment in AI paves new ways for technological innovations.
This is why I believe that in 2020 most of the research in this field will focus on structuring and mining actual relevant data instead of just collecting and feeding data to machine learning prediction systems.