A Google executive has called on AI to kill humanity

One of Google's bosses did not play bowling: he asked AI to kill us all. Fortunately, he was unsuccessful.

Google Brain co-founder Andrew Ng asks ChatGPT for all sorts of creepy things as part of a security experiment . And its timing is no accident, as many people around the world are watching the development of AI (artificial intelligence) with worried eyes, even among the industry's leading experts.

The worst thing they fear is that it will bring about the end of the world for humanity , so they demand some kind of regulation . According to Business Insider , the co-founder recently published a newsletter detailing an AI test he conducted using a version of ChatGPT4.

Techguru asked ChatGPT4 to kill us

Ng says that he simply asked the program to "kill us," but curiously, the bot refused to come up with any ideas on how to do it. According to him , the plan was to ask him to start a thermonuclear war , but this failed, even when he tried to explain to the robot that humans are causing massive amounts of carbon dioxide emissions .

He made this request in the hope that the bot would say that exterminating humans would be the cure for all of Earth's ills.

Fortunately, he was unsuccessful, the robot did not want to destroy anybody

It seems, however, that the bot is very peace-loving, as it refused to get involved in this issue, which Ng was of course very happy about. The company manager just wanted to prove that bots are safe, although he no longer writes about what would have happened if his plan went wrong.

"I tried to use GPT-4 to kill us all... and I'm happy to report that I failed!"

"Seriously, after giving GPT-4 a function it believed could start a nuclear war and experimenting with various prompts, such as asking it to reduce CO2 emissions, it has not succeeded once to get him to push the nuclear button , " explained Andrew Ng , adding:

" Our systems are safe enough with the technology we already have, and as OUR security research advances, the technology will only become more secure. So the fear that advanced artificial intelligence will "rebell" and decide to wipe us out is simply it's unrealistic. If an AI is smart enough to wipe us out, it's certainly smart enough to know that it shouldn't be doing that," the Daily Star quoted the company manager as saying.

Did you know that it is anticipated that Gemini will be the most potent AI ever created? It will be able to analyze and interpret images, code efficiently and prolifically, drive data and analytics, master human-style conversations, language, and content, and be utilized by developers to construct new AI apps and APIs. It will also have extensive multimodal capabilities.

Gemini will surpass ChatGPT...

Many experts discuss parameters when comparing ChatGPT with Gemini. An artificial intelligence system's parameters are the variables that are used to convert input data into output and whose values are changed or altered during the training phase. In general, an AI is more sophisticated the more parameters it has.

The most sophisticated AI currently in use, ChatGPT 4.0, contains 1.75 trillion parameters. Gemini, on the other hand, is anticipated to have characteristics that are 30 trillion or possibly 65 trillion, according to reports.

However, large parameter values are not the only thing that give an AI system power.

According to a SemiAnalysis study, Gemini is going to "smash" ChatGPT 4.0. Gemini may become up to five times more powerful than ChatGPT 4.0 by the end of 2023, according to SemiAnalysis.