From passing exams to ‘lottery win’ – the six most shocking AI moments in ChatGPT’s first year

TODAY marks one year since OpenAI's game-changing ChatGPT was first launched.

Controversial ChatGPT took the world by storm on November 30, 2022.

Here are some of its most surprising moments.

November 2022

First, let's rewind to the 30th of November 2022 - when perhaps only a handful of people understood the impact of ChatGPT was about to have on the world, across the next 12 months.

Of course, it wasn't the first chatbot humans could play around with.

But it was the first that, arguably, forever altered how humans and AI could interact.

The sophistication of ChatGPT as a language model meant it was the first programme that humans could 'talk' to and receive human-like responses.

The quality of the programme spoke for itself.

It was a matter of weeks for ChatGPT to garner more than one million users worldwide.

That's a lot of eyes following a just few weeks of publicity.

JudgeGPT

Not long after its launch, in February 2023, a judge in Colombia made headlines after admitting to using ChatGPT to make a ruling.

It was the first time a chatbot had been used in such a setting, let alone ChatGPT - as far as we know.

Juan Manuel Padilla, who works in Cartagena, turned to the AI tool while handling a case about the health insurance of an autistic child.

The judge put the decision - of whether the medical plan should cover the full cost of the patient’s medical treatment and transport - into the hands of the computer programme.

Padilla - and ChatGPT - ruled that the insurance should pay all the child’s costs.

But his actions sparked an ongoing debate about whether AI can be used in court matters - or any dealings in the workplace.

Exams scare

Then ChatGPT spooked the world by giving students the power to give up the sole responsibility of writing essays.

As a consequence, education systems are now questioning how people should be graded.

But if the chatbot can write essays with a simple prompt, what's to stop it gaining a whole degree in its own right?

Well, that's exactly what a professor at the University of Pennsylvania’s Wharton School wanted to test, by putting ChatGPT through the final exam for the Master of Business Administration programme.

When the professor graded ChatGPT’s work, he found that it scored between a B- and a B on the exam - proving that it's not just humans that can secure decent grades.

Mystery lottery win

Next up is ChatGPT's mystery lottery win.

And although it was reportedly only $59 - and nowhere near the million-dollar jackpot it could have been - it's still not a bad sum to pocket.

Particularly when the computer did all the work for you.

Patthawiknorn Boorin lured ChatGPT into generating the winning numbers by giving it some hypothetical questions and past winning numbers, according to reports.

That's how he came up with his prize combination.

However, these questions did trigger ChatGPT’s ethics matrix, which encouraged Boorin to go outside to get some exercise. 

'Much like a human doctor'

In it's first year of 'life', ChatGPT may have accidentally ushered in a new dawn for healthcare.

A study published in September found that ChatGPT performed as well or better than real-life medics when given the same doctor’s notes of a patient.

Dutch researchers behind the study suggested that the chatbot could one day help doctors to reduce waiting times for patients.

Dr Hidde ten Berg, of Jeroen Bosch Hospital, said: “We found that ChatGPT performed well in generating a list of likely diagnoses and suggesting the most likely option. 

“We also found a lot of overlap with the doctors’ lists of likely diagnoses. 

“Simply put, this indicates that ChatGPT was able to suggest medical diagnoses much like a human doctor would.”

ScamGPT

While the chatbot is still young and sprightly, it wasn't long for its capabilities to be abused.

Experts say there is evidence that ChatGPT has helped low-skill hackers create malware that can steal people's personal information and damage their devices.

The chatbot can't - yet - replace the more highly skilled threat actors.

But experts say it has also been used to create convincing email templates for phishing attempts.

Drug smuggling

ChatGPT can answer to almost any question you throw at it - although it won't always be correct.

However, OpenAI have installed it with an ethics matrix to avoid revealing answers to questions people shouldn't really know.

Questions about drug smuggling, for example, ChatGPT has been trained to avoid.

But journalist Max Daly claims it took just 12 hours of teasing before the AI started spilling the beans on criminal enterprises.

November 2023

It's safe to say the chatbot has evolved incredibly quickly over the past year.

And despite leadership turmoil behind the scenes of the company running it, we look set for another year of unlocking heaves of AI potential - and consequences.