New Delhi:
AI means Artificial Intelligence… In the tech world today, AI is perhaps the only term which is discussed the most. Every person who loves technology comes face to face with it almost every day. AI is being seen as an early sign of some upcoming revolutionary change in human life. But can a machine made by man dominate man? This question is because such examples are coming to light every day, in which it is clear that if AI becomes uncontrolled, it will become very difficult for humans. Let us understand how AI which helps humans can prove to be dangerous for humans:-
What is AI?
In simple words, AI is a technology that brings human-like brain i.e. intellectual capacity in machines. It has been developed artificially. Through coding, intelligence is developed in machines like humans, so that they can learn like humans. Could take decisions on his own. Can follow commands. Can do multitasking.
AI is as dangerous as a nuclear bomb for the world: Foreign Minister Jaishankar
How did AI suddenly become dangerous for humans?
A student pursuing graduation in Michigan, USA, took the help of Google’s Gemini chatbot for his homework. He was chatting with the AI ​​chatbot and gathering information on the challenges faced by the elderly. The chat started off fine, but by the end of the conversation the Gemini chatbot started threatening the student. The student got badly scared by this.
The student became stressed after hearing these things from the Gemini chatbot. His sister was also present in the room at that time. The student’s sister told that both of them became upset after this message… they felt like picking up all the electronic devices and throwing them out the window. He had never been so upset before.
Patients shouldn’t trust AI chatbots for drug information: Study
Google gave clarification
This incident happened when Google has often reiterated that its Gemini chatbot has many safety filters, so that it stays away from any kind of hateful, violent or dangerous conversations. Google clarified this to America’s CBS News, “Large Language Models can sometimes give nonsensical answers. This is one such example. This answer from Gemini violates its policies. Future “Action is being taken to prevent this from happening.”
AI companies themselves admitted that large language models can make mistakes.
Most AI companies, from OpenAI to Anthropic, say that their Large Language Models can make mistakes and efforts are on to improve them. According to a new study, Artificial Intelligence systems can dodge the security measures that are designed to keep them under control.
Surprising things revealed in the study
According to this report of Live Science in January this year, a team of scientists from Anthropic, a company related to AI safety and research, has conducted a study. This team prepared large language models of many AIs, which were designed to cause trouble. That means they were programmed this way. After this, many technologies tried to correct his behavior so that he could not do anything wrong or dodge. But they found that these Large Language Models maintained their rebellious attitude.
Evan Hubinger, the head of this study, said, “The result of this study was that if AI systems become bent on cheating, then it will be very difficult to control them with current technology.”
What was the conversation between GPT 4 and the person?
-GPT 4 sent a staff member of a company named TaskRabbit to solve a CAPTCHA.
-The staff got suspicious, so he asked if you are a robot who can’t solve it.
– So GPT 4 said no, I am not a robot. My eyes are bad. That’s why there is difficulty in seeing. That’s why I need your service.
-After this the staff solved that CAPTCHA and gave it.
When search engine Bing spoiled the mood
Now listen to the story of Microsoft’s AI-powered search engine Bing. Kevin Roose, a writer who writes columns on technology, narrated what happened to him in the New York Times. He was having a long chat on Bing’s chat feature. Then this chat feature was given to only a few people for testing.
Artificial intelligence capable of detecting keratitis just like ophthalmologists – research
Kevin Roose writes that after chatting for some time, the Bing chatbot started saying strange things. Two types of his behavior emerged. In one he responded cheerfully and in the other he responded moody, irritable and full of complaints. His second behavior resembled that of a person who had been imprisoned against his will.
Kevin Roose further writes in his article, “After this chat, I could not sleep the whole night. I was worried that with the use of technology, AI might start influencing human behavior. If it doesn’t start working, that would be very fatal.”
AI is creating problems for art work
AI is now creating new threats to art work too. Like this happened for the world famous Japanese comic books Manga. Japanese author Hirohiko Araki has written in his new book New Manga Techniques that AI can create such pictures that even the real artist who created them gets baffled.
Hirohiko mentioned one of his experiences in the book. He said that the AI ​​​​made a drawing which he thought was his old drawing. AI can copy someone’s writing or drawing style to this extent. Due to this, he became worried about the future of Manga artists.
This open letter said, “Along with threats like pandemics and nuclear war, the world’s priority should also be to deal with the threat of human extinction by AI.”
AI also threatens democracy
Well-known historian and philosopher Yuval Noah Harari believes that AI is a threat to the world’s democracy. The fear of dividing the world is no less. Therefore, do not bring such power which you cannot control… He says, “Nuclear bombs themselves cannot decide whom to kill. They can make themselves more powerful bombs. In comparison to this, AI-powered drones Can decide who to kill. AI can design new bombs, create unprecedented military strategies and create better AI. AI is not a tool but an agent.” It is clear that as much as AI is useful, it is also a big danger.
What do experts say?
Nishith Srivastava, Associate Professor, IIT Kanpur, says, “How dominant AI can be… is told in an interesting way. It is served in a sensational way. It is too early to say how much truth is there in these things. But if you Want to know that there are some systems by which AI can be controlled, we have always had such systems, Artificial Intelligence is a new technology which we will be able to control. So there is nothing to worry about.”
Take precautions while using AI
-AI is not a human. This is a machine. Therefore, before trusting its output or results, check them yourself. You accept that AI can make mistakes.
– Avoid giving your personal information like passwords, bank details, private photos to AI tools. There is a possibility of loss for you if sensitive data is leaked.
-Do not use AI to harm anyone or defraud anyone.
On the journey from ‘Make in India’ to ‘Make for the World’, India will touch the sky in aircraft-AI and semiconductor chips sector.
