.png)
Leave us a message
We usually respond within a few hours. You can also make an appointment with us for a specific date by clicking on the link to our calendar below.
In the issue of 'The Economist' of 12 June 2020, an interesting technological addition appeared, the main thesis of which is: Is the 'autumn' of artificial intelligence coming?
Artificial Intelligence, which has been made public again since about 2010, is a concept almost as old as the computer and has had its ups and downs for decades.
Alan Turing wrote his first chess programme in 1948, three years after the creation of ENIAC, considered to be the first computer in history.
The term "artificial intelligence" ("artificial intelligence") appeared in the scientific literature as early as in the 1950s, and the first chatbot (ELIZA psychoanalyst) in 1966.
There was a wave of enthusiasm for rapid progress both in the 1970s and 1980s, and then the subject waited for a renewed surge of interest until the 'cloud revolution'. The entry of the 'cloud' has increased the computing capacity enormously and thus made it possible to analyse dependencies in huge datasets to a greater extent.
Artificial intelligence and data science have been called 'new electricity'. Why, then, do you fear that artificial intelligence will develop further?
The main argument is that the results obtained by artificial intelligence algorithms are as good as the quality of the data on which the algorithm works.
An interesting example of this is the unmanned Amazon shops (only present in the United States so far) - Amazon Go.
Shops do not have any cash registers, either manned or self-service. The shop operates on the basis of a network of cameras, sensors that transmit data to an application based on AI algorithms and a mobile application that the consumer scans at the entrance and exit of the shop.
Unfortunately, in the first phase of the project, it turned out that life creates much more scenarios than the Amazon application analyzing images from cameras. People can take the product, put it away, take it back in their hands, put it in a bag, then take it out, etc. The most problems were reportedly caused by the movements of small children shopping with their parents :)
In turn, the application used by Amazon to help recruits select CVs for software engineers favoured men, as most of the candidates in the database were male.
Similarly, IBM found out that the AI recruitment application favours white people - there was a shortage of black people in relation to the overall proportion in society.
In 2018, Mount Sinai Hospital in New York recorded that the application of the X-ray scanning of pneumonia was less accurate in terms of images other than those from the hospital where it was 'trained'.
What can be done in this situation? One of the answers to the problem of lack of quality data is for companies using AI applications to produce artificial data that serve AI applications in the learning process. This is the way Amazon Go went. He created another application that creates graphical data that imitates the image from the camera, on which data the application analyzing the data learns.
According to 'The Economist', teaching applications using artificial data is a trend that will intensify.
Another trend highlighted by the report "The Economist" is the concentration of AI applications on the so-called "AI boring". Instead of focusing on ambitious projects related to medical research or even autonomous vehicles, the use of AI will shift to less ambitious, but easier to implement with today's state of the art projects, e.g. management of supply process support robots.
Personally, it seems to me that the calculation possibilities offered by the 'cloud' will not reduce but rather increase the importance of AI. Nevertheless, I recommend this report to everyone.
And what do you think about this?
+48 600 911 085
We usually respond within a few hours. You can also make an appointment with us for a specific date by clicking on the link to our calendar below.