Can we trust AI?

    Can we trust AI? 

Artificial intelligence is a tool that doesn't create information from emptiness. That means the AI uses data and information that already exists. If something or somebody changes information from the data sources that the AI uses, the AI gives wrong answers. The AI can use old-fashioned information. In the same way, a human can take an old book of facts from the shelf. 

That is one thing that we must remember. The second thing is that most AI chatbots are owned by companies whose purpose is to bring profits for their owners. That causes a situation where some of the answers that the AI gives are customer-friendly. That means they are made to please customers, or people who pay for those chatbot services. 

However, one of the most important aspects of using AI is that it is actually utilized incorrectly. But the question is, why is it used? Because it makes work more effective. That means people who work with the AI assistants have no time to think and check the answers that the AI gives. The prime focus in AI use is that it makes people do more work. 

The modern business ecosystem has evolved over time, when all products were physical. That means the business ecosystem measures the effectiveness of the worker by simply calculating the number of work performances in a time unit. And that means the number of code lines that a programmer makes determines how effective that person is. And that means there is no time to analyze and think about the product that the AI makes. 

This is the problem with AI. It makes people more effective, and it should leave time to make analyses about the product. But the problem is that if the AI allows the person to work at double speed, that gives the company leaders the opportunity to fire half the workers. That decreases costs. 

When we talk about how AI makes things less effectively than a top-level programmer, we can find one very interesting question. If we analyze that argument closely, we face the claim that all programmers are top-level specialists. All programmers are top-level specialists. And another thing is that some workplaces are so hurried that they have no time to advise junior programmers. Sometimes old workers think that the young comrade is a competitor, and that makes the workplace atmosphere poisonous. 

The person might not dare to ask for advice from workmates. And that causes firing. The thing is that the AI is a tool that can help people. But we must remember that this tool allows us to analyze and think about new possibilities, only if we give it that chance. But we must have time and willingness to check the data that the AI gives. And if we see the AI only as a tool that allows us to cut the personnel costs, that is not the way to make our working life easier. 

If we just copy-paste texts that the AI makes. And don't even try to analyze that data, we don't increase trustworthiness in the company. That means we deliver the decisions to the AI. AI can also use old-fashioned datasets. Or it can use fake or falsified data. That means the AI makes mistakes in the same way as humans. It simply takes the wrong book from the shelf, and then that causes destruction or catastrophe. 

The AI is not immune to fake information. The problem with propaganda home pages is that their page rank are raised by drumming it with net queries. That makes those homepages easier to find. The problem with those pages is this: they use .com or some other neutral terminal identifiers. So, those homepages are not easy to connect to China or Russia. They are operated from servers that are in Western countries. The operators use a remote control to control those servers that operate in Western sockets. 

If the AI uses regular web browsing applications, that means it can select a propaganda page. The popular pages are always at the top of the page lists. And that thing means that the AI selects popular homepages. That decreases the AI's trustworthiness. The AI requires training. That training means that the users must set limits on what kind of sources those systems use.