All the developments in the field of artificial intelligence could create and deploy digital minds which no one can understand, not even their creators. This fear has been expressed in an open letter published on the website of Future of Life Institute. There is a call to AI developers to hit the pause button on training any models more powerful than ChatGPT. This call is from experts and researchers in this field. More than 1100 people in the industry signed a petition calling the labs to stop training powerful AI systems for a period of at least six months. Among the signatories are Elon Musk and Steve Wozniak, the Apple co-founder.
Powerful AI systems should be developed only when there is an assurance that the effects will be positive and the risks will be manageable.
In recent past, AI has demonstrated the ability to generate text and art. GPT4 is even more powerful. It can cut jokes and could pass tests.
Policy makers and AI developers must work together to formulate governance systems and create oversight bodies.
In the petition, the signatories have requested the governments to intervene if major players in the AI field do not agree to a verifiable pause.
AI should refocus on making todays systems more robust, accurate, transparent and trustworthy.
The signatories also include Youshua Bengio and Emad Mostaque. There is no signatory from OpenAI, the company which has developed ChatGPT and GPT4.
Sam Altman of OpenAI, the company that created ChatGPT negotiated with Microsoft in 2019 a billion dollar investment to help OpenAI develop artificial general intelligence (AGI) enabling the machine to do what a human brain could do. Altman put the project on par with the Manhattan Project that dealt with the US effort to build the atomic bomb during World War ll. Altman believed that artificial general intelligence project would bring prosperity and wealth like no one had ever seen. Of course, he was worried about the harm it could cause even to the extent of destroying the world as we know it.
All this happened in 2019. In 2023, people have realised that Altman was more prescient than others.
ChatGPT is now available at the click online-a technology that can answer many queries about nuclear physics and organic chemistry, and can write a beautiful essay on London of Past and Present and can create a computers programme that drops digital snowflakes on the laptop screen.
However, the same technology can spread falsehoods, can persuade people to do what they should not. The last of week of March witnessed a petition of AI experts to put a moratorium on AGI technology.