The Sci-Fi portrayals of AI feature machines that are able to think just like humans end up dominating the world – and typically with devastating outcomes. Let’s see how that stacks up against where AI is today.
CAN DO: AI large language models (LLM) can now be trained to do incredibly complex tasks. Those capabilities are continuing to increase.
To continue reading, please login
Not yet a member? Find out about joining
CAN’T DO: Where humans still have an edge is “sentience,” the ability to have subjective experiences. To feel. To taste. To experience pain.
DEBATE: There are factions within the scientific community that believe that AI is already sentient.
IN REALITY: “An LLM is a mathematical model coded on silicon chips. There is a profound difference between how humans generate sequences of words and how an LLM generates those same sequences.
“When I say ‘I am hungry,’ I am reporting on my sensed physiological states. When an LLM generates the sequence “I am hungry,” it is simply generating the most probable completion of the sequence of words in its current prompt,” per Stanford University professors Fei-Fei Li and John Etchemendy.
INSIGHTS: While AI will likely replace a lot of tasks, people will still be needed for ‘impact evaluations’. “There are types of human evaluations that are still extremely useful and can tell us things which we cannot learn any other way,” per Professor Ehud Reiter, Chair in Computing Science, University of Aberdeen.
