What We Need to Know About AI

“Artificial intelligence (AI) is the science of teaching machines to learn human-like capabilities. Artificial General Intelligence (AGI) is the point at which AI can perform all human cognitive skills better than the smartest humans.” And another quote for you: “The Coming Wave: An emerging cluster of related technologies centred on AI whose transformative application will both empower humankind and present unprecedented risks."

Like most people of my age (I suspect), I have heard the term ‘Artificial Intelligence’ and immediately switched off. Too big, too complex, too complicated, too technological and too late in my life for me to get my head round. Or so I thought until very recently.

Then I heard someone called Mustafa Suleyman speaking on the podcast ‘The Rest is Politics: Leading’, and was hooked by what he was saying. Suleyman has a compelling backstory. He was born in Islington in north London to a Syrian-born taxi driver and an English nurse. Despite an unpromising start in life, he studied philosophy at Oxford but left (ultimately) to start DeepMind, which he sold to Google in 2014 for £400m at the age of 30. Suleyman’s book ‘The Coming Wave’ was published on September 5th and I have been devouring it since it arrived a few days ago. It’s beautifully written, very easy to understand, even for someone like me, and I thought I’d share some of his insights about an aspect of fast developing technology which you may have heard of called ChatGPT.

The book opens with a potted history of other significant and important waves of technological advancement. “In the space of around a hundred years, successive waves took humanity from candles and horse carts to one of power stations and space stations.” By the way, a ‘technology’ is any scientific advance that helps you to do something either faster, more efficiently or more cheaply. A simple example would be a pair of spectacles, a technology which helps you to see more clearly. An advanced technology would be the sunglasses worn by Chinese police with an inbuilt facial recognition device capable of tracking suspects in crowds. If that sounds far-fetched and somewhat scary, I can assure you that it may be troubling, but it’s certainly not far-fetched. It is happening right now. This is why we need to pay attention to AI. There’s much to like and even to welcome, and there’s also much to concern us.

It’s also about the here and now, not the future. And I quote: “Mass scale AI rollout is already well underway. Everywhere you look, software has eaten the world, opening the path for collecting and analysing vast amounts of data. That data is now being used to teach AI systems to create more efficient and more accurate products in every area of our lives [...] A raft of applications rely on techniques that a decade ago were impossible. [...] AI systems run retail warehouses, suggest how to write emails or what songs you may like, detect fraud, write stories, diagnose rare conditions and simulate the impact of climate change. They feature in shops, schools, hospitals, offices, courts and homes. You already interact many times a day with AI; soon it will happen much more, and almost everywhere it will make experiences more efficient, faster, more useful and frictionless.”

Sounds like some kind of futuristic utopia doesn’t it? But, as Suleyman says at the start of his book, the spread and accessibility of these systems also poses significant risks. So I thought I’d tell you about ChatGPT and the rise of Large Language Models (LLMs) under the headings of the good, the bad and the ugly (as I interpret it).

The Good.  It was thought that processing natural language was too complex for modern AI. Then in November 2022, OpenAI released ChatGPT. Within a week it had more than a million users who loved that if you asked it a question it replied instantly in fluent prose. Ask it to write an essay, a syllabus for a physics course, a dieting manual, a Monty Python script, a press release, a business plan in the style (say) of the Bible or a 1980s rapper, it will do so in seconds. And how on earth is that possible?

A big part of human intelligence is that we use past experience to predict the future. Intelligence may therefore be seen as the ability to generate a range of plausible scenarios about what may happen and then base sensible actions on those predictions. This is how the development of Large Language Models started in 2017 at Google. “LLMs read very large numbers of sentences, learn an abstract representation of the information contained within them, and then, based on this, generate a prediction about what should come next.” These systems are called transformers, so GPT stands for Generative Pre-trained Transformers. Over the period 2019 to the present day, GPT has developed from GPT2 which had 1.5 billion parameters (which are the measure of an AI system’s scale and complexity), to GPT4, released in March 2023, which has 1.7 trillion parameters and can now also work with images and code. It appears to ‘understand’ spatial and causal reasoning, medicine, law and human psychology’. 

All of this is just the start. As Suleyman says “we are only beginning to scratch at the profound impact LLMs and ChatGPT are about to have. They are the first sign of a wave beginning to crash around us. In 1996, thirty-six million people used the internet; this year it will be well over five billion. That’s the kind of trajectory we should expect from these tools, only much faster. Over the next few years AI will become as ubiquitous as the internet itself: just as available, and yet even more consequential”.  It has the potential to make human beings more effective, and more efficient and save us a lot of time and money. 

The Bad. Especially if you are a screenwriter currently on strike in Hollywood whose pay has actually dropped over the past 10 years and for all those whose whole livelihood may be at risk from the development of ChatGPT. The latest LLMs are “stunningly good at scores of different writing tasks once the preserve of skilled human experts, from translation to accurate summaries in writing plans for improving the performance of LLMs”. As the technology becomes more robust and reliable and more widely available, the negative impacts will be felt, especially amongst those whose occupations will inevitably be threatened because AI will be able to do their work much faster and much more cheaply. 

The Ugly. I’m not sure how I’ll feel about watching a play created by ChatGPT by ‘AI Shakespeare’ as opposed to all those familiar ones penned by our greatest living playwright five hundred years ago. I have a horrible feeling that if it entertains me, stimulates my imagination and is thoroughly enjoyable, I may not mind as much as I should. But I am truly terrified by the dystopian prospect of AI-created videos of real people apparently saying and doing things that they have never said or done. The technology to do this is available now and In the hands of malign actors and used to (say) manipulate electoral outcomes, it is a prospect that I find truly terrifying. 

Until very recently I had zero interest in this subject. It felt too vast, too ‘techy’ and too unfathomable to engage with. But, isn’t it interesting that when you start to focus on something which you’ve been ignoring, suddenly it’s everywhere? Since I started to write this, two articles in the Financial Times have popped up in my news feed in quick succession. One called ‘Generative AI exists because of the Transformer’ explaining how Large Language Models are created, and the other ‘The Global Race to Set the Rules for AI.’ And then just last night on the ITV programme ‘Peston’ one of the guests was Mustafa Suleyman himself talking about the book ‘The Coming Wave’, which I have been discussing here. He was very impressive. Articulate, calm and surprisingly reassuring.

Perhaps you, too, have zero interest in this subject, but that won’t make it go away. Just recently my local supermarket opened about 8 more automatic checkouts, leaving a single regular one ‘manned’ by a human being. How long before there are no checkouts at all and we are automatically charged for everything we place in our baskets (as currently happens in Amazon Go stores). 

As we age we can become dinosaurs, like someone I know who still adamantly refuses to have a mobile phone, or we can become ostriches, pushing our heads further into the sand in the hope that the world as we know it will stay the same until we’re gone. If Suleyman is right about the coming wave of AI, then this is a vain hope. I prefer to believe that ignorance is not bliss and that knowledge is power. And at least I can have a conversation with my grandchildren and have a vague notion of what they are talking about!!

Tricia x

Watch Our Latest Video...

Upcoming Events:

Friday 29th September

Monday 2nd October