Productive AI systems like ChatGPT, DALL-E, and Midjourney are poised to revolutionize countless industries. Artificial intelligence expert Pascal Kaufmann he reported to ZDNet that these new AIs are so good at generating text and images that it's very easy to overestimate their capabilities.
Kaufmann, founder of Starmind.ai and Mindfire Group explained that true intelligence is defined as the ability to solve entirely new problems. While it's easy enough to be impressed by AI's ability to process natural language, Kaufmann argued that large language models like GPT-3 simply mimic intelligence without actually possessing it. In particular, they lack the ability to face challenges they have not faced before.
For example, these systems can create poetry and write code because they have been trained on existing content. But they cannot have abstract tasks, such as understanding humor or having empathy.
“There are some people who treat the term 'AI' as an 'amazing innovation'. So whenever something is amazing, it's called AI," Kaufmann said.
“But when you look at it from a scientific perspective, you can measure and define intelligence. Intelligence can be measured as an ability to deal with invisible problems, problems that are not encountered. Can you solve something that is completely new? Large language models like ChatGPT are unable to deal with anything they haven't encountered before. Therefore, his intelligence is zero.”
Despite this limitation, Kaufmann acknowledged that the new models are a significant advance in AI capability and could revolutionize several industries. It highlighted their ability to produce creative work efficiently, significantly reducing costs and implementation times. He also suggested that models can generate more creative solutions to unique problems than humans, thanks to their access to a vast pool of information.
Kaufmann pointed to the value of artificial intelligence in industries such as illustration and film production, where virtual human beings can be created that will be much more economical than real-life humans.