Mit dem FORBES-NEWSLETTER bekommen sie regelmässig die spannendsten Artikel sowie Eventankündigungen direkt in Ihr E-mail-Postfach geliefert.
A couple of minutes ago, my colleagues once again tried to ask Alexa about the pandemic. The result was predictable. For the umpteenth time, I was told that artificial intelligence is not smart at all, and that I, as a programmer, should be ashamed of the industry. This is not the first time I have been confronted about the state of artificial intelligence and asked about the prospects for that industry. The global artificial intelligence market size was valued at USD 27.23 billion in 2019 and is projected to reach USD 266.92 billion by 2027. However, according to the Morning Consult report published in 2020, only 16 per cent of consumers said they would prefer to be assisted by a voice system, while 71 per cent on average said the same of human help. This makes the question of the future of AI quite controversial.
At Evrone, we have been doing custom development for many years, but, as a programmer, I do not usually answer questions about the future of AI. Instead, I give the floor to my neurophysiologist colleagues. Modern science does not yet understand how the brain organizes the mind, but they suspect that the secret lies somewhere in the learning process.
We programmers have learned to do something similar to what neural networks do in the brain. However, neurophysiologists do not have a complete understanding of how the learning of individual brain cells and neurons, leads to the learning of the whole brain, which has a hundred billion such cells, all connected in a certain way.
When programmers write about AI in magazines and articles, they usually mean "an artificial neural network trained to solve an isolated problem." This type of problem or task can be something like image recognition or assembling a phrase based on a simple rule and a million existing phrases referenced from literature. However, the brain learns as a whole, and it learns simultaneously, unlike artificial neural networks that learn individual tasks separately. A newborn child looks for patterns in an avalanche of information and commits them to memory, a process that the Russian scientist Pyotr Anokhin calls "cognite", and the American scientist Michael Graziano calls "Attention Schema.”
...is the co-founder of Evrone, a software design and development consultancy. Entrusted by the Fortune 500 companies like KFC, L'Oreal, Evrone design award-winning digital products and critical business IT solutions for enterprise clients and startups.
In my opinion, the key difference between artificial intelligence and the intelligence of the human brain is in how they learn. Philosophers say, "We are what we have learned." A child perceives the world around them through their body. They are surrounded by other people, with similar bodies, who are trying to convey to them knowledge about the world around them. The child's brain can look for patterns between how they see their body, control it, how they see the bodies of other people and interact with them. We shouldn’t forget that the connection of the parts of the brain and its topology have adapted, over millions of years of evolution, for precisely this kind of learning, through the interaction of bodies.
The huge tree of associations in the brain of an adult contains connections between millions of meanings, from simple ones concerning light and shadow, which result in our perception of visual illusions, to the advanced connections that give us the ability to speak, think about the complex world, and apply logic.
Therefore, although we call it "artificial intelligence," a more accurate description would be "artificial fragments of the brain." Until the systems we create have a semblance of a body and the ability to learn through interaction with society, we are unlikely to be able to do business with Alexa or ask her advice on life issues.
Text: Anthony Cherepanov
Opinions expressed by Forbes Contributors are their own.