OpenAI has released its much-anticipated GPT-5 model, which it claims is a major improvement on its GPT-4 predecessor
While the new AI model is more user friendly, the advancement is more evolutionary than revolutionary, according to an article in MIT Technology Review
In a demonstration prior to the launch, Yann Dubois, a lead at OpenAI, asked GPT-5 to create a web application that would help his partner learn French
When the same prompt was given to GPT-4, it produced an identical app which was less aesthetically pleasing
However, OpenAI’s GPT-5 is faster, more efficient and less likely to make incorrect claims, known as ‘hallucinations’
The creation of more reliable AI is a crucial step towards artificial general intelligence (AGI) and superalignment.