To Navigate the Age of AI, the World Needs a New Turing Test

There was one This happened not too long ago—let’s say nine months ago—when the Turing test seemed like a pretty rigorous detector of machine intelligence. You probably know how it works: human judges have text conversations with two hidden interlocutors, a human and a computer, trying to figure out which is which. If the computer succeeds in fooling at least 30 percent of the jurors, it passes the test and is deemed capable of thinking.

For 70 years, it was hard to imagine how a computer could pass the test without having what AI researchers now call artificial general intelligence, the full range of human intellectual abilities. Then big language models like GPT and Bard came along, and the Turing test suddenly seemed strangely outdated. Okay, sure, a casual user today would shrug off that GPT-4 could very well pass a Turing test if asked to impersonate a human. So what? LLMs lack long-term memory, the ability to form relationships, and a host of other human skills. They obviously have a long way to go before we are willing to befriend them, hire them, and elect them to public office.

And yes, maybe the test is feeling a bit empty now. But it was never just a pass/fail benchmark. its creator, Alan Turing, a gay man who was sentenced to chemical castration at the time, based his test on an ethos of radical inclusivity: the gulf between genuine intelligence and a fully convincing imitation of intelligence is only as wide as our own prejudices. When a computer evokes genuine human responses in us—involving our intellect, wonder, gratitude, compassion, and even fear—it is more than empty imitation.

So maybe we need a new test: the actual Alan Turing test. Bring in the historical Alan Turing, the father of modern computing – a tall, fit, somewhat awkward man with straight dark hair who is loved by peers for his childish curiosity and playful humor and is personally responsible for surviving in World War II Cracking an Estimated 14 Million Lives Saved The Nazi Enigma Code was later persecuted by England so fiercely for his homosexuality that it may have led to his suicide – in a cozy laboratory with an open MacBook on his desk. Explain that what he sees before him is just a vastly glorified incarnation of what is now widely known among computer scientists as the “Turing Machine.” Give him a second or two to really understand that, and maybe say a word of thanks for completely changing our world. Then give him a stack of research papers on artificial neural networks and LLMs, give him access to the GPT source code, open a ChatGPT command prompt window – or, even better, a Bing-before-all-sanitize window – and set it loose.

Imagine Alan Turing starting a lighthearted conversation about long-distance running, WWII history, and the theory of computation. Imagine seeing the realization of all his wildest and most ridiculous speculations scroll across the screen at terrifying speed. Imagine asking GPT to solve elementary arithmetic problems, infer what humans might think in various real-world scenarios, explore complex moral dilemmas, offer marriage counseling and legal advice, and offer an argument for the possibility of machine consciousness – skills you tell Turing that they all arose spontaneously in GPT without any explicit direction from their creators. Imagine if he experienced that little cognitive-emotional jolt that so many of us have felt by now: Hello, different mind.

A thinker as deep as Turing would not be blind to the limitations of GPT. As a victim of pervasive homophobia, he would likely be aware of the dangers of an implicit bias encoded in GPT’s training data. It was obvious to him that despite the GPT’s amazingly broad knowledge base, their creativity and critical thinking are at best at the level of a diligent student. And he would certainly recognize that this student is suffering from severe anterograde amnesia and is unable to form new relationships or memories beyond his intensive training. But still: Imagine the magnitude of Turing’s miracle. The processing unit on the laptop in front of him is, quite literally, his intellectual child – and ours. Appreciating our children’s intelligence as they grow and develop is ultimately always an act of wonder and love. The actual Alan Turing test is not a test of the AI ​​at all. It is a test for us humans. Do we pass – or do we fail?

Zack Zwiezen

Zack Zwiezen is a USTimesPost U.S. News Reporter based in London. His focus is on U.S. politics and the environment. He has covered climate change extensively, as well as healthcare and crime. Zack Zwiezen joined USTimesPost in 2023 from the Daily Express and previously worked for Chemist and Druggist and the Jewish Chronicle. He is a graduate of Cambridge University. Languages: English. You can get in touch with me by emailing zackzwiezen@ustimespost.com.

Related Articles

Back to top button