The need for transparency is one of the most pressing issues of modern AI, which can remove a lot of misunderstanding and mistrust
Technopedia, February 2, 2021
AI technologies are becoming more common in our daily life. That is why, when delegating AI responsibility for our health, our safety on the road or decisions in court, it is necessary to have full understanding and clarity about how these smart technologies work.
On the one hand, after a number of critical cases in which AI made rather controversial decisions, the demand for the transparency of AI work has increased. On the other hand, a technology such as deep learning is very difficult for human perception in its structure, with its “black box” nature – and all because the work in such systems occurs as a result of the interaction of millions of computing units that are difficult to track. The existence of such a black box AI and the need for its transparency that has spawned a new discipline in AI known as explainable AI (XAI). The concept of XAI is to make predictions made by artificial intelligence clear and understandable without compromising performance.
Along with protecting against biases and improving AI trustworthiness, the XAI can make smart systems more robust against adversarial attacks. This is possible due to the fact that, within the framework of the concept, the adversarial input will be easily detected, since the system will explain the reasons for its abnormal behavior, thereby issuing an attack.
Despite the fact that this trend is only developing, such a concept of open system operation can significantly facilitate the solution of security issues and raise confidence in the work of AI systems.
Forbes, January 27, 2021
The number of smart appliances in our lives is expected to increase in the next few years. In order for us to use comfort in our daily life, we need to be confident in their safety and security. Several European and US companies and associations have already begun to work on drafting laws regulating safety and security aspects of ethical smart systems.
However, there is still a lot of work to be done in this direction, because smart systems need a truly thought-out structure and support when it comes to making safe and ethical decisions. In addition, it is also important to consider physical security issues in case something goes wrong. Modern trustworthy AI and ML systems have thoughtful development and design, and also know how to correctly and ethically interact with people. And of course, while paying attention to the ethical side of smart systems, we must not forget about the security aspect, since such systems are still vulnerable to hackers; the security issue should begin to be addressed at the stage of both hardware and software development.
My Tech Decisions, January 13, 2021
The FBI recently warned of the increasing incidence of so-called “swatting attacks”.By itself, swatting is a term which denotes a false appeal to the emergency services. For example, an imminent threat to the life of a person may be reported in order to elicit an immediate law enforcement response.
In this very case, attackers used stolen email data to enter smart devices and then perform malicious calls. It all started when one of the manufacturers of smart home devices reported a number of attacks to law enforcement agencies: attackers hacked into devices of this company, which had smart audio and video functions.
It was also clarified that such attacks can be motivated by revenge, a form of persecution or just prank, but from the point of view of the law, this is a serious crime that can have fatal consequences for other people. One way or another, this situation once again highlighted the importance of protecting seemingly harmless home smart devices.
The development of new technologies requires a lot of responsibility, because any great force can be both good and evil Averting a cyber pandemic: how businesses are building a global ...
