Standard Digital Edition

So, could ChatGPT do your job better than you?

The first AI language bot for public use is making waves by successfully mimicking human beings — but despite the fears, says Rohan Silva, it may prove a positive tool

I TYPED in the question — “explain what ChatGPT is in the style of a Cockney” — and within a few seconds the computer-generated reply came gushing out: “’Ello gov’na! Let me tell ya’ about ChatGPT, the bleedin’ AI language model wot can talk the talk and walk the walk. See, ChatGPT is a powerful machine wot’s been trained to understand and generate human language.

“But watch out, ’cause like any powerful tool, ChatGPT can be used for good or for bad. So, use it wise and fair, or ya’ might find yerself walkin’ the plank!”

It’s a response which, I’m sure you’ll agree, started off convincingly Danny Dyer, but ended more Jack Sparrow.

While it’s obviously not perfect, you can understand why ChatGPT is sparking so much debate. You can ask it to write a sonnet about Sadiq Khan (“So let us honour London’s noble son, who works to make his city, shining, one…”) or a one-line summary of Prince Harry’s new book (“A personal and raw account of his mental health struggles...”).

There have even been news reports that ChatGPT-generated answers would have passed the entrance test for the prestigious Wharton MBA course in the US. It’s a level of sophistication that has many worried but others excited — including Microsoft who on Tuesday announced a multi-billion-dollar investment deal with OpenAI, the company which created ChatGPT, to try to get ahead of its competitors in the artificial intelligence arms race. OpenAI, which counts Elon Musk among its early investors, is now valued at nearly $30 billion — quite an achievement for a company with only 350 employees.

ChatGPT uses a “deep learning model”, basically crunching through vast quantities of text on the internet and using that data to figure out how to construct human-sounding sentences.

Meta’s chief AI scientist has claimed

that ChatGPT is “not particularly innovative”, because its underlying technology is not that different from what’s being developed at other companies.

Whether that’s true or not, there’s no denying that ChatGPT is one of the most tangible uses so far of artificial intelligence — the term for software capable of doing tasks that previously only humans could do (such as writing complex prose), and potentially outsmarting us too (like at chess, where no human can beat the latest AI machines).

At the same time as ChatGPT, we’ve also seen the emergence of AI software like DALL-E (also from OpenAI) and Midjourney which can generate amazing images and artwork in seconds — the kind of thing that until recently only humans could have created. I’ve been writing for years about the potential impact of AI on jobs and society — and how London is at the forefront of this shift, thanks to British companies like DeepMind, Faculty and Stability AI.

In the same way that automated factories reshaped the labour market by replacing many “blue-collar” jobs with robots, some experts predict that AI will displace large numbers of “white-collar” roles. The Bank of England thinks up to 15 million jobs could be lost in the UK, with administrative and clerical roles being among the most at risk because they can be most easily be replaced with advanced software. And it’s not hard to think about other potential downsides, like AI-generated fake news, or students cheating by getting ChatGPT to write their homework. But here’s why I think the AI future may be brighter than the pessimists believe. For many workers, AI software will be a tool that ends up helping, not replacing us.

JUST look at how Microsoft is reportedly incorporating ChatGPT into Word and PowerPoint — meaning that your computer will increasingly be able to do the grunt work of writing and putting together presentations, as long as you point it in the right direction.

Google is looking to do the same with Gmail, so that it can write emails for you — and they’re apparently planning chatbot for their search engine to make it easier to find information online.

Another good example is teaching, where techies are experimenting to see how ChatGPT could be used to mark students’ work. If these approaches come to pass, they could help free people from some of the most boring and time-consuming parts of their jobs — and give us more space to focus on what’s meaningful and important.

But the main reason to be positive is that AI could unlock innovations that we couldn’t come up with by ourselves. Perhaps the best case study is here in London, where DeepMind researchers have made astonishing breakthroughs in working out how proteins — the building blocks of life — fold themselves into different shapes. If this problem can be cracked, it could lead to new treatments for diseases such as Alzheimer’s and Parkinson’s, and potentially enable scientists to develop new sources of clean energy.

DeepMind’s software has been sorting through huge amounts of data and spotting patterns far faster than a person could — but this raw AI computing power has also been combined with the brainpower of humans, who have used the software’s findings to develop new insights and breakthroughs.

Demis Hassabis, co-founder of DeepMind, says using AI in this way “allows us to be much more creative, to try many more ideas, often in parallel”. Instead of replacing scientists, it complemented their skills.

We’ve already seen ChatGPT listed as a co-author on academic papers — with physicist Kai Polsterer describing AI systems as “hardworking assistants” that can leave you free “to do the cool, interesting science on your own”.

What we’re starting to find in science may end up being a better prediction of the future than the vision of a world where robots destroy all our jobs.

It’s a working life with the potential to be far more interesting than we might have imagined — and more creative too.

I’ll leave the last word to ChatGPT, which gave this response when I asked about the years ahead: “AI can be both good and bad, depending on how it is designed, developed and used.”

Sounds suspiciously like a smarmy politician’s answer — but it’s probably close to the truth.

The future may be brighter than the pessimists believe: for many, AI will be a tool that ends up helping not replacing us

Upfront

en-gb

2023-01-26T08:00:00.0000000Z

2023-01-26T08:00:00.0000000Z

https://eveningstandard.pressreader.com/article/281668259108034

Evening Standard Limited