“I Am Not a Robot” | By Demi Wait
“A computer would deserve to be called intelligent if it could deceive a human into believing that it was human.”
Those are the words of mathematician Alan Turing, who created what he called “the Imitation Game” (now known as the Turing Test) to determine whether a computer can truly “think”. There’s quite a visible divide in opinion on AI, where some people love it and use it in their day-to-day activities, and others are perhaps more hesitant to use it. Then there’s always the horror film scenario of some kind of robot uprising, where AI comes to take our jobs and ruin our lives. I have no idea if this is what Turing was envisaging when he created the Turing test.
There have been instances where this test has been passed. It’s often argued when the first pass was. Some would argue that it was as long ago as 1966, though a more recent pass that made the news was Google's AI chatbot, LaMDA. Some are now suggesting that the Turing Test actually measures a machine’s deceptive ability, rather than intelligence.
So, are people right to fear the “rise of AI”? Could AI ever be smarter than humans?
Usually, people with these fears are picturing something closer to Artificial General Intelligence, or AGI. Essentially, it’s AI with a human level of functionality. At the minute, AGI is just a concept and has yet to be created. It’s basically a form of AI that could perform a range of tasks that require real human intelligence.
Machines can’t function like humans yet. Think of how many times you’ve clicked that button, “I am not a robot”. Or you’ve selected the squares that do have traffic lights in them. It’s not that a robot can’t press the “I am not a robot” button, but more about how the mouse moves.
A machine would move to it in a straight line, at a consistent speed. I’d be impressed if we could all do that. And so, we as humans are allowed in when we move the mouse inconsistently!
Also, there have been many cases of a machine not being able to recognise an image properly in certain contexts, hence the captcha tests that look more like this:
Just because we’re not at AGI, doesn’t mean that it’s not affecting people as it is now. While AI isn’t always good at image recognition, AI image generation has taken over the internet lately. Generative AI can be incredibly useful for people in reducing the time taken to complete mundane tasks, like creating schedules.
Most people don’t tend to have a problem with using AI to help them complete these small tasks – it’s not replacing them! However, it can be harmful to artists, who spend hours on their work for a machine to emulate something similar in a matter of seconds. There are frequent questions over what art and images AI is trained on to be able to generate images. There have been discussions surrounding how OpenAI have been secretive about the data they trained their model with, though errors have been found in ChatGPT where it will accidentally reveal confidential training data. It’s been seen that chunks of data training data have been exposed when ChatGPT is asked to repeat a number or phrase infinitely. Companies will likely need to be far more transparent regarding their training data before they’re even close to truly being able to replace people.
All in all, while AI is rapidly advancing, I think it’s pretty safe to say that humans and AI are pretty distinguishable from each other at the minute. AI does have its upsides and downsides, but AGI is still just a theory at the minute.
Can you tell that I wrote this? I’ll clarify that I did not use ChatGPT to write this. I am not a robot.
Demi Wait is an Advanced Data Fellowship apprentice at Multiverse based in London, England, and is writing for The Apprentice Lens. Here's more about her:
"Hi, I'm Demi. I currently work as an Apprentice Data Scientist for Colt. Outside of work, I love writing of all sorts and am most likely found with a book in my hand. I find huge inspiration in all sorts of tech, real-world or fictional!"