Today the impossible has become a fact. Technology catches up with us in terms of image analysis, linguistic communication, and even ... creating art or virtual reality. Having a photo of a person, you can create a video from this one photo, and with an additional recording of their voice, a specialist is able to build a speech synthesizer and make him say with his own voice things that he never said - says Kamil Konikiewicz, director of the Artificial Intelligence Development Office at PKO BP in an interview with PAP.
This year's Covid-19 pandemic has significantly accelerated the digitization process. Intelligent solutions in the form of "learning" algorithms used in computer programs or in smartphone applications have become commonplace. We introduce AI (Artificial Intelligence) into our lives using social networking sites, streaming platforms, installing various devices and consenting to the processing of data about us.
As Kamil Konikiewicz explains, there is no fully autonomous AI so far, which means that algorithms are not able to learn without human supervision. Someone has to tell them that this particular photo has a dog or that the conversation is about coronavirus. Data must be properly prepared, categorized and described. Only after delivering a sufficiently large sample of them, the algorithm begins to learn.
“The skills that algorithms acquire in point application are amazing. For example, they can describe better than people what they see in a given picture, and even create art - by >> painting << pictures or composing pieces of classical music. They are also great at games - today, teaching them, for example, to play chess, we place two algorithms in front of each other, we load them the rules of the game, they start to play with each other and within a few hours using the so-called amplified learning reaches grandmaster level, ”explains Konikiewicz.
The director of the Artificial Intelligence Development Office at PKO BP is a co-creator of the voice assistant in the IKO banking mobile application. The assistant uses advanced analytics and artificial intelligence algorithms to enable users to deal with banking matters using voice commands, e.g. transfer, phone top-up, search for transactions in the account history, enter the BLIK code.
The system understands natural language and does not dictate the method of communication. You can communicate with him in a completely unhindered way. Over 100,000 have already been used to teach it. phrases, also colloquial. Checking the account balance can be ordered for over 2,000. ways, and make the transfer by saying, for example, "push a hundred to Janek". Importantly, the list of phrases that the algorithm will understand is much larger, you can talk to him in a completely natural way.
However, banking applications operate under a very strict regime of Polish and European law imposing requirements as to secure transactions. Each login and important transaction must be confirmed by a strong authorization method - entering a PIN or entering an authorization code. It is also possible to set the fingerprint or face recognition functions.
"We moved a lot of our lives online, and cybercriminals followed us there," Konikiewicz points out. - As a large bank, we and our clients are exposed to various types of attacks or threats, while the degree of our security is very high - not only at the level of applications and systems but also at the level of security of an e-mail account or business phone. We also have very dynamic security and cybersecurity departments, ”he assures.
As the director of the AI Development Office at PKO BP points out, "artificial intelligence interferes in our lives as much as we allow it". “People can reveal information that they shouldn't, even sensitive data when talking to virtual assistants. Designers of this type of solutions should protect these data, sometimes even with the "mouth" of the bot from disclosing them when it is not required, ”emphasizes Konikiewicz.
In his opinion, the virtual world brings great opportunities, but also certain threats. The popularization of social media and smartphones (including digital smartwatches or bands) causes that they collect more and more data about users and their environment. "False information, sound and visual materials are gigantic challenges for which we are not yet civilized ready," emphasizes Konikiewicz.
The exponential development of this field is a huge challenge faced by the institutions regulating the issues of artificial intelligence - i.e. the number of applications is so enormous that the creation of a universal law regulating all possible variants of AI applications is extremely difficult.
For program and application developers, the challenge is the availability of data (in the case of a voice assistant for specific solutions for the Polish language) and privacy regulations.
“It's hardest to regulate and oversee big global organizations like GAFA (Google, Amazon, Facebook, Apple), so you turn a blind eye to the grey economy. If, for example, advertisements appear in a popular internet messenger in the context of what I said or saw during a video chat with my friend - then something is wrong because I do not remember allowing me to analyze a video stream or audio stream from private talks ”- notes Konikiewicz.
"The greatest world powers are investing billions of dollars in the development of AI. Who will conquer this market - will conquer a stronger geopolitical position in the future - he adds. - We have a lot of intelligent, talented people who know this technology and it would be good to make use of it - build development centres, education centres and use intellectual potential to prepare society for the responsible use of such technology "- he points out.
In the expert's opinion, it would be useful for the technology related to artificial intelligence to develop primarily in the public sector - in particular in supporting diagnostics in health care, security, cybersecurity, in solving social problems such as domestic violence, unemployment, as well as in monitoring protection environment.
The development of technology is also a need for high-class specialists: Data Scientists, AI engineers, programmers, UX designers, and in the case of voice assistants, also linguists.
According to Konikiewicz, initiatives are needed that will make the so-called "average citizen" will know more about AI. Admittedly, in Poland, there are timid attempts to introduce educational and investment programs or pro-Publico Bono foundation activities aimed at educating citizens. However, the rapprochement of the scientific and business sectors as regards the Polish market is still too early in the development and popularization of this technology.
Konikiewicz also points to the need to introduce the subject "digital media" at the level of primary school, during which children and young people would learn about the possibilities of modern technologies, but also about the dangers of their improper use.
"The material would be enough for several years of teaching" - emphasizes the expert of PKO Bank Polski.
Source of information: PAP MediaRoom