This column first appeared in The Western Mail in December 2022.
In a break from the usual family tradition, we won’t be playing Trivial Pursuit this festive season. Instead, you’ll find us hunched over a laptop experimenting with our latest obsession – ChatGPT. I hope I’m not overstating it to say this mind-blowing technology feels like a seismic shift in how we use the internet, and I’m already addicted. So what exactly is it? Concentrate – here comes the geeky bit.
The easiest way to explain ChatGPT is that it’s an incredibly sophisticated chatbot. If you’re wondering what that is, if you’ve ever communicated with a bank or retailer online, then it’s likely you’ve already used a chatbot. Unfortunately, they’re usually unimpressive to the point of making you want to throw things at your screen and demand to speak to a human. But ChatGPT is a million times better than any chatbot you’ve ever used. It uses powerful artificial intelligence-driven neural networks to turn questions and statements into highly accessible, human-like language.
Think of it as a mix between Google and an incredibly powerful Siri or Alexa.
Or, as my teenage son put it – “A reason never to do my homework again”.
If any of his teachers are reading, we’ve had words about that. But his point is a good one. Because ChatGPT can do a whole range of things that blow my mind every time I log in. It can answer any question you can think of, compose songs and poetry, write emails and essays – and best of all, it SEEMS to provide answers that are as good as anything the human mind could come up with.
The launch of the chatbot, created by Microsoft-backed company OpenAI, has led to widespread concerns that students will use the software to cheat on assignments. As a result, academics have already suggested that schools and universities develop new ways of assessing students in response to AI’s threat to academic integrity.
We tested it by asking it to write 1000 words on Paris during the Belle Epoque, and the essay it churned out in real-time in front of our eyes was astonishingly accurate – and made for compelling, natural copy. I asked it to suggest Christmas presents for a vegan, and it came up with some genuinely good ideas that would never have occurred to me. A friend (and mischievous wag) who shall remain nameless asked it to write a poem about an act that I daren’t repeat in a family newspaper; her disappointment when it returned the message “This content may violate our content policy” was priceless. But it was also reassuring, given that teenagers worldwide are signing up for accounts. Over one million people have signed up in a week since the platform was launched – a rate of adoption that took Instagram many months to achieve.
Confession time. I did consider asking the software to draft my column this month, but a pang of deeply ingrained Catholic guilt prevented me from doing so. But the potential impact of this technology on a whole range of industries is impossible to ignore. From journalism to the legal profession, it’s possible to imagine a future where human beings – with their pesky sickness absence and pay demands – become entirely dispensable as machines become increasingly adept at producing articles, contracts and a whole range of content. The potential impact on the employment market is clear – and how do we replace the jobs that will inevitably be lost?
Cardiff West MP Kevin Brennan has raised concerns in Parliament about the impact artificial intelligence and the data mining of creative work could have on the creative sectors because its capabilities could lead to AI creating facsimiles of music, songs, art and books without paying the original artists or creators a penny. It’s another example of technology moving faster than our legislators’ ability to keep up.
I believe this technology will revolutionise how we interact with the internet. If Google was a massive step forward in the late 90s in terms of the way we organise information, ChatGPT is the equivalent for 2022 and beyond. And, as with all technological advances, this brings with it serious risks.
Operating in a secretive manner much like a ‘black box’, the AI draws on a big bucket of information which comes out as naturalised language. This means we lose the ability to fact-check or verify anything. Chat GPT doesn’t so much scrape information as generate new information based on undisclosed and uncited sources. With Google, you can judge the credibility of the sources presented to you, but because ChatGPT doesn’t reveal its sources, it becomes much more difficult to sift misinformation. Considering the Brexit and Cambridge Analytica scandals, the social implications of this are terrifying if you think too hard about it.
The question is, will our policymakers be able to move quickly enough to mitigate these risks? And where will the technology go next, and how quickly?
As I ponder these big questions, I’m also busy asking it to do completely frivolous things purely for the LOLs.
So here’s a special Christmas Haiku to all you lovely Western Mail readers, from ChatGPT:
Snowflakes falling gently,
Silent night, holy night, peace reigns.
Merry Christmas all.
I couldn’t have put it better myself.
Over and out.
Leave a Reply