The Turing Test

​

         Even before modern computers as we know them existed, computer scientists and philosophers have been interested in knowing whether machines will one day be able to think. As a result, creating a thinking machine has become an important goal in the modern field of artificial intelligence. When the question of thinking machines is explored, Alan Turing and his contributions to the field of computer science take center stage. Turing, an English mathematician and computer scientist, published the influential paper “Computing Machinery and Intelligence” in 1950. The paper has formed a foundation for computing as we know it today, and it explores Turing’s view on thinking machines.

​

         Turing first prioritizes having clear definitions and a question in “relatively unambiguous words.” Turing thinks that the question “Can machines think?” is not worthy of discussion because the terms are not rigorously defined. Instead, he poses a question related to it, in the form of the imitation game. He claims that any machine that could do well in his imitation game could be considered a thinking machine.

​

         Turing defines the imitation game as a game played by three players, one person, one machine, and one interrogator. The interrogator is separated from the person and the machine but can communicate with both through text. The goal of the interrogator is to ask questions to both the player and the machine to determine which is which. The goal of the machine is to fool the interrogator into thinking that it is human, while the goal of the person is to ensure that the interrogator correctly identifies the machine. Both accomplish their goals by cleverly answering the interrogator’s questions. Turing claims that the best strategy for the machine is to give answers that would replicate the behavior of a human. Turing also rigorously defines the machine in the game as a digital computer. His idea for this digital computer is now named a Turing machine in recognition of the importance of his contribution, as this Turing machine is equivalent in computational power to any computer that can be created today.

​

         Turing predicts in his 1950 paper that he envisions computers will one day be able to pass his test, so we can likely consider them thinking machines. Turing also predicts that the idea of a machine thinking will eventually be normalized as well. However, he recognizes the possible objections to his argument, and he refutes many of them in the original paper.

​

For example, the first argument, the argument from theology, argues that it is the sway of some higher force that gives a body the ability to think and that the body and mind are separate. Turing refutes this argument by asking why God could not simply grant a soul to any entity, organic or inorganic. Another argument regards the mathematical limitations that Turing realizes any machine will have. From this chain of logic, there ought to exist certain questions that no machine will ever be able to answer but that a human could answer. Asking a question like that should prevent any machine from ever being successful in the Turing test. Turing refutes this by asking if there is any reason to believe that the human mind is unbounded in comparison to the machine or if there are any questions that a human would never be able to answer.

​

         Another famous argument Turing replies to is Lady Lovelace’s objection. Ada Lovelace, another English mathematician famous for her work on Charles Babbage’s analytical engine, remarks that a machine could never create anything new. Since machines are programmed to do what we tell them to do, a machine could never do anything that we did not already know how to do. Turing refutes Lovelace’s objection with the argument that in a deterministic universe, nothing is new. Every new action is just a series of reactions between previous actions.

​

         Modern philosophers and computer scientists have continued to analyze Turing’s arguments and build upon them. For example, philosopher John Searle’s Chinese Room argument that an entity could simulate intelligence without being intelligent. If a computer is just following instructions, can we consider it to be thinking or even having a form of mental state? Searle describes a scenario where an English speaker follows instructions to convert English words to Chinese symbols. The English speaker could be successful at this task, but we could not consider the human to have learned Chinese. Searle argues that the computer performs a similar function to the English speaker in this scenario, so we would not consider the computer to be thinking.

​

         Other philosophers have also proposed alternatives that build on Turing’s test that will set good goals for the development of artificial intelligence technology. However, the question of machine intelligence is a tricky one, and while Turing made advancement far ahead of this time, we have a long way to go. It is likely that there will never be one all-encompassing test for measuring machine intelligence, but researches can set realistic, achievable, and meaningful goals to advance our understanding of artificial intelligence.