Why cant artificial intelligence solve everything?

Artificial intelligence systems require a lot of data to get started, but the public sector usually does not have the appropriate data infrastructure to support advanced automatic learning. Most of the data remains stored in the offline archives. The few digital sources of data tend to be buried in the bureaucracy. Often times, the data is published across different government departments and each requires special permissions to be accessed. On top of all this, the public sector usually lacks human talent with the right technical capabilities to fully reap the benefits of robot intelligence. For these reasons, the excitement about artificial intelligence has attracted many critics. Professor in computer science at Berkeley University, "Stuart Russell", has always called for a more realistic view focusing on simple everyday applications of artificial intelligence rather than virtual control by super-intelligent robots. Likewise, a professor of robotics at the Massachusetts Institute of Technology "Rodney Brooks" writes: "It takes almost all of the innovations in the field of robotics and artificial intelligence to spread more widely than people in the field or outside of it imagine." One of the many difficulties in deploying AI systems is that they are very vulnerable to hostile attacks. This means that malicious AI can target another AI to force it to make false expectations or behave in a certain way. Many researchers have warned against launching artificial intelligence without appropriate security standards and defensive mechanisms. However, AI security remains an often neglected topic. If we are to reap the benefits and reduce the potential harm of artificial intelligence, then we must start thinking about how to apply automatic learning in a meaningful way to specific areas of government, business and society. This means that we need to have a discussion about the ethics of artificial intelligence and the suspicion many people have about automatic learning. And most importantly, it is necessary to be aware of its limits, people still need to take the lead. Instead of drawing an unrealistic image of the power of artificial intelligence, its important to step back and separate his actual technical potential from magic. For a long time, Facebook believed that problems, such as the problem of spreading misinformation and hate speech, could be identified algorithmically and stopped. But under recent pressure from lawmakers, the company quickly pledged to replace its algorithms with an army of 10,000 human references. The medical profession has also realized that artificial intelligence cannot be considered a solution to all problems. The oncology program "IBM Watson" was part of artificial intelligence, which was intended to help doctors treat cancer. Although it was developed to provide the best recommendations, human experts have found it difficult to trust the machine. As a result, this program was abandoned in most hospitals where it was tested. Similar problems arose in the legal sphere when algorithms were used in courts in the USA to try criminals. An algorithm calculated the results of the risk assessment and advised the judges on the judgment. The system was found to exist to duplicate structural racial discrimination and was subsequently abandoned. These examples show that AI does not have a solution to everything. Simply using artificial intelligence for artificial intelligence itself may not be productive and always beneficial. Not the best solution to every problem is by applying an AI to it. This is the critical lesson for everyone aiming to boost investments in national AI programs: all solutions come with a cost and not everything that can be automated should be.