Would the events featured in the Terminator movie series play out?
A team at OpenAI, the company that spearheaded the rise of artificial intelligence (AI) with ChatGPT, has now created a team to make sure that their machines will never work against human beings. According to an official announcement from the company, they are building a team to "steer and control AI systems much smarter than us."
Superintelligent AIs can arrive within the next decade, OpenAI estimates. These machines could help with most if not all the world’s problems with its advanced mechanisms. However, its coding is not to be trifled with, should it arrive in our daily lives. The company warns users that these programs can be very dangerous, and could lead to human extinction in the worst possible scenario.
The new team, led by the OpenAI’s research lab’s current chief scientist and head of alignment, as well as OpenAI’s top researchers and engineers and current compute power, aims to create a "roughly human-level automated alignment researcher." This is a proposed AI system that will not go outside set parameters. As to how they would do this, ZD Net’s Artie Beaty cites three important steps:
1. Develop an understanding of how AI analyzes other AI without human interaction
2. Use AI to search for any problem areas or exploits
3. Deliberately mistrain some AI to see if that gets detected
Read more about their new project here.
Image credit: olia danilevich