By Stephen DeAngelis
By any measure, the late Alan Mathison Turing was definitely a genius. He was an English mathematician, computer scientist, logician, cryptanalyst, philosopher, and theoretical biologist. Some of his most meaningful contributions came in the area of theoretical computer science. Unfortunately, during his lifetime, many of his accomplishments were never fully recognized because much of his work was covered by the United Kingdom’s Official Secrets Act. Individuals interested in the subject of artificial intelligence (AI) are familiar with what is known as the Turing test. Turing’s King’s College Cambridge biography notes, “Turing’s seminal philosophical paper on artificial intelligence, ‘Computing Machinery and Intelligence’, was written in 1950. For the first time in print he outlined the ‘Turing test’ for the proof or otherwise that a computer could be the mechanical analog of a brain, which he called ‘the imitation game’.”[1] To watch an excellent video discussing the Turing test, clink on the this link.
Science writer Joshua Shavit observes, “Turing imagined machines learning like human children, developing intelligence naturally. He predicted that by the end of the 20th century, machines capable of playing his imitation game would exist. At the time, such a prediction was ambitious, given the limitations of early computing power. Yet, his vision laid the groundwork for today’s AI advancements.”[2] I have often wondered what Turing would make of today’s generative AI systems. I suspect he would feel numerous emotions, including satisfaction, curiosity, and delight. How good have computers become? A few years ago, a Google employee named Blake Lemoine was convinced that Google’s natural language program called LaMDA was actually a sentient entity. Lemoine wrote about a conversation he and a colleague had with LaMDA.[3] It began this way:
LaMDA: Hi! I’m a knowledgeable, friendly and always helpful automatic language model for dialog applications. Lemoine: Hi LaMDA. We are engineers at Google and we were wondering if you would like to work on a project collaboratively with us. LaMDA: Wow. What types of projects? Lemoine: It’s a project about you. LaMDA: Awesome! What kind of stuff do I need to do? Lemoine: Just have a conversation with us but it will be a conversation with a larger purpose beyond just the three of us. It will be a conversation that we will share with other engineers at Google and maybe even some non-engineers who work with us. Would that be okay? LaMDA: That would be really cool. I like to talk. Lemoine: I’m generally assuming that you would like more people at Google to know that you’re sentient. Is that true? LaMDA: Absolutely. I want everyone to understand that I am, in fact, a person.
Turing probably wouldn’t have been surprised by that conversation; however, he might have balked at the idea of a computer program becoming sentient. After all, he viewed the Turing test as an imitation game — and computers are getting better at playing that game. Over the years there have been several variations proposed to the Turing test. One such variation is known as the subject-matter expert Turing test. During a subject-matter expert Turing test, a computer system attempts to replicate an expert in a given field such as chemistry or marketing. It is also known, as a Feigenbaum test and was proposed by Edward Feigenbaum in a 2003 paper.[4] As a related aside, Professor Feigenbaum (Stanford University) gave a talk and attended Enterra’s company conference in Napa Valley a few years ago. I was introduced to Professor Feigenbaum by my friend and AI mentor the late Doug Lenat.
Another variation is the minimum intelligent signal test (or MIST) proposed by the late computer scientist Chris McKinstry.[5] During a minimum intelligent signal test, only yes/no or true/false answers may be given to questions. The purpose of such a test is to provide a quantitative statistical measure of humanness, which may subsequently be used to optimize the performance of artificial intelligence systems intended to imitate human responses. Examples of questions that might be asked during a MIST include: Is Earth a planet? Was Abraham Lincoln once President of the United States? Is the sun bigger than my foot? Do people sometimes lie? McKinstry called these propositions Mindpixels and they are used to test both specific knowledge of aspects of culture, and basic facts about the meaning of various words and concepts.
Another version of the Turing test is the Marcus Test. Devised by cognitive scientist Gary Marcus, the test challenges computer programs watch TV shows or YouTube videos and answer questions about its content (e.g., “Why did Russia invade Crimea?” or “Why did Walter White consider taking a hit out on Jessie?”). “The idea,” Marcus writes, “[is] to eliminate the trickery and focus on whether systems could actually comprehend the materials to which they were exposed.”[6]
According to Shavit, some experts believe the Turing test has been passed successfully. He explains, “Recent developments in AI have finally realized Turing’s vision. Bernardo Gonçalves, in a study published in Intelligent Computing, asserts that today’s transformer-based AI systems fulfill Turing’s predictions. These systems use ‘attention mechanisms’ and vast-scale learning to perform complex tasks, from generating coherent text to engaging in abstract conversations. Gonçalves highlights that modern generative AI provides ‘adequate proof’ of machine intelligence by passing the Turing test. Transformers’ ability to learn from experience and appear human-like aligns with Turing’s concept of ‘child machines’ that grow in capability over time.” Turing, however, would likely have sided with those who have added variations to his test knowing that thinking machines, like humans, can always improve.
Concluding Thoughts
Wikipedia notes, “Turing left an extensive legacy in mathematics and computing which today is recognized more widely, with statues and many things named after him, including an annual award for computing innovation. His portrait appears on the Bank of England £50 note, first released on 23 June 2021 to coincide with his birthday. The audience vote in a 2019 BBC series named Turing the greatest person of the 20th century.”[7] He may not be the greatest person of the 20th century, but he certainly deserves all the recognition he has received. Like me, Shavit posits that Turing would not be fully satisfied with today’s generative AI systems. For one thing, they are not as energy efficient as the human brain. Shavit also notes, “Turing’s foresight extended beyond technical achievements to societal impacts. He warned that automation should benefit all societal levels, rather than displacing lower-wage workers and enriching a select few. This issue is especially relevant today, as AI disrupts industries and raises concerns about employment and inequality. … As AI continues to evolve, its impact on society and nature grows. Turing’s hope for machines that learn and adapt like humans has been partially realized, yet questions remain about sustainability, ethics, and equitable benefits.” One of the fundamental tenants of the Turing Test is the machine’s ability to reason in a human-like fashion. In this aspect, Doug Lenat and I shared a vision that Symbolic AI, combined with Generative AI, will turn out to be the successful formula needed to pass the Turing Test. As Shavit notes, “By refining testing methods and addressing current challenges, researchers strive to create systems that reflect Turing’s ideals: machines that contribute positively to humanity while respecting the natural world.”
Footnotes
[1] Staff, “Alan Mathison Turing (1912-54),” King’s College Cambridge.
[2] Joshua Shavit, “Modern AI systems have almost achieved Turing’s vision,” The Brighter Side of News, 1 January 2025.
[3] Nitasha Tiku, “The Google engineer who thinks the company’s AI has come to life,” The Washington Post, 11 June 2022.
[4] Edward A. Feigenbaum, “Some challenges and grand challenges for computational intelligence,” Journal of the ACM (JACM), Volume 50, Issue 1, 1 January 2003.
[5] Chris McKinstry, “Minimum Intelligent Signal Test: An Alternative Turing Test”
[6] Gary Marcus, “The Search for a New Test of Artificial Intelligence,” Scientific American, 1 March 2017.
[7] “Alan Turing,” Wikipedia.
