Member-only story
GPT-4 could solve a 60-year old Challenge in Computer Science
Cooking is a common analogy to programming. You have to write down instructions in a very clear way as the computer just executes those instructions — not more, not less.
We are not used to make such precise instructions, which is hilariously shown in the “exact instructions challenge”:
Computers, in their current state, do not possess the ability to think; they merely execute explicit instructions. However, the potential productivity gains offered by genuine thinking machines that can comprehend the nuances of human communication and compensate for imprecise or incomplete orders are immense. The concept of thinking machines has long captivated the human imagination, with philosopher René Descartes musing about automata interacting with humans as early as 1637. Yet, it was only with the advent of modern computers that the notion of intelligent machines began to solidify.
Fast forward to 2023, and the development of ChatGPT has made the prospect of thinking machines increasingly tangible. This progress invites a critical question: At what point can we consider a machine to be truly thinking?
The Turing Test: An imitation Game
The computer scientist Alan Turing published “Computing Machinery and Intelligence” in 1950 in which he describes an imitation game. Shortened to the relevant part it goes like this:
You have a human interrogator, a human player, and a computer player. The interrogator may talk with both, the human player and the computer player over a chat interface. The interrogator must figure out who the human is.
The core idea was that this would be a sure sign of intelligence. If you cannot distinguish a human from a computer anymore, how could you say it is not intelligent?
