Thursday, December 22, 2022

The AI called Chat GPT caused panic in big companies like Google.



The lack of flexibility is a problem in large corporations. When the corporation is big people believe that the entire business environment dances around the massive corporation. The dominating position in the marketing area or marketing sector makes the company blind. 

And that causes an effect where those corporations neglect their R&D process. Actors like Google believed that their search engine and other products cause situations where some new actor will overtake them. The New York Times tells there is some kind of panic in Google. That company's AI development is behind the point where it should be. 

AI is the next-generation tool. That system will destroy the internet in the form we know it. And the thing that makes AI so powerful is fuzzy logic. Fuzzy logic means that if the person gives a command that is not clear, the AI can ask: "would you repeat the command". Or it can send the list of possibilities of what that spoken command could mean. Then the system asks: was the command in the list?

If the system uses traditional or precise logic. The missing end of the word causes the system stuck. So, there must be something that denies the stuck. 

But fuzzy logic means that the AI can try to fix the problem. That kind of thing is more effective if people are using mobile interfaces. The new mobile interfaces require a new type of user interface. The new mobile interfaces should allow people to walk and look straight ahead when they use their systems. 

The new mobile interfaces are hybrid systems that are connected the mobile telephone to HUD glasses and voice command systems. The system can use voice commands and give the information to the HUD screen. But in this case, if that environment is noisy, the person is tired or gets winded the system might not recognize the words. 

*********************************************************'

The fuzzy logic means that the system would not use precise data. If a person gives the command "what is the highest speed of "trai.."?. The system selects the words that are involving the letters "trai", and then it can fill the end by using the letters "...er" or "...n".  So the AI can ask would the user mean "train" or "trainer"? 

********************************************************

The system can also use EEG from the Wernicke area that is producing speech. But if the system is unable to detect words it must ask what the person meant. Then the system can calculate the blinks of eyes to select the word that the user meant. That kind of interface is needed in mobile systems that are used when the person is walking or running. 

If we want to find information by using the AI-based system the AI can simply, collect data from multiple sources. And then create a new homepage or essay-type document. 

The weakness of the AI is that it sorts the home pages and then selects paragraphs by using certain orders. It doesn't understand words that it searches. That means it searches words like the "find" button in the right corner of the web browser. 

The AI can simply calculate the number of words that it finds from certain paragraphs. Then it selects the paragraphs where the most wanted words are. Then the AI must be careful that the same paragraph is not chosen again. And the AI can select the first paragraph from the first page because there is the possibility that paragraph is the most suitable for the beginning of the text than some other paragraph. 

And one of the orders is the position of those homepages in the list. So the AI selects the first paragraph from the first homepage, the second from the second, etc. And that causes the situation. AI makes many mistakes if it sorts the home pages by using the URL. The number of sources must be limited. Or the answer will be very long. 

But more complicated AI can search the certain word from the list. And that makes the result more suitable. The ability to use information makes AI more effective. But also how much, and how accurate information the AI can use plays a big role in that kind of system. 

When we want to develop AI there should be people who will develop the search algorithms. The development would happen by using academics or others who are estimating the thesis to estimate the results that AI can give. Then they can give feedback about the results that the AI makes. And that data can use to create mode-accurate AI-based technology. 

The fact is that the development of the net is a very fast-moving process. And traditional actors must step away if they cannot flex and develop solutions that are suitable for commercial use. The thing that makes Chat GPT powerful and effective is that everybody can see the power of AI. 



No comments:

Post a Comment

Note: Only a member of this blog may post a comment.

The String theory and the 5-plet problem.

"Concept image of strange particles in an atom." (InterestingEngineering) The 5-plet is a strange 5-particle group detected in the...