Way back in the very different world of 1968, the film 2001: A Space Odyssey presented a villain like no other, Hal 9000. The doorbell-sized camera showed a potential future where artificial intelligence was cunning, emotive and downright evil.
We’re now 22 years on from the setting of this futuristic film and while the past year has been rampant with developments in AI, it isn’t as threatening as the years of science fiction would have had us believe.
Instead, the futuristic tool that most people will be familiar with from the world of artificial intelligence is ChatGPT. A platform that takes worded prompts and returns… well, whatever you ask for.
It can create poems, write essays, offer ideas for new companies, have a conversation with you and just about everything inbetween. But while ChatGPT is evidently an impressive development, it isn’t without its clear flaws.
The model is often confused, can be caught off-guard with clever prompts and isn’t winning any awards for the creativity of its poems. However, this is all starting to change as the technology evolves.
OpenAI, the company behind ChatGPT, has now announced the latest version of the technology known as GPT-4. This gives the chatbot the ability to deal with far longer reams of text, better understand complex ideas, and even express creativity to a higher level.
What it means above all is that the tool that started off writing bad poems, confused articles and painful jokes is getting smarter. It can do more, understand more, and the thing that will worry most people: challenge more industries.
In early examples of GPT-4, OpenAI demonstrated its ability to process images, creating a full working website from an image of a drawn website plan. It could do a tax return and even analyse the food in your fridge to offer up potential recipes.
Now seemingly more than just a way to create viral content, what does the future of ChatGPT hold? “There are no end of applications for this technology because it is really a gateway to our own knowledge,” explains Peter Bentley, a computer scientist and author based at University College London.
“It makes existing knowledge accessible efficiently. Early and very basic applications were using it write documents, or blogs. Then we discovered it could write computer code.”
While these early applications still had uses in the real world, they were limited to the likes of writing news articles and dealing with smaller tasks around copy, like writing terms and conditions or a website’s manifesto.
“Now it can understand images so it could be used to summarise the content of images, or make complex suggestions from images (e.g describe how this building could have its energy efficiency improved or tell me which trousers would go with my shoes). This may be the new way we search the internet,” says Bentley.
A new way of using the internet is something we hear a lot (we’re looking at you, metaverse), but ChatGPT is already there. Microsoft has implanted the technology into its Bing search engine and is currently using the updated GPT-4 version.
Equally, Google is now working on a competitor called ‘Bard’. It would work in a similar way to ChatGPT and realistically, would be used to improve Google’s search engine in the same way as Microsoft has with Bing.
“It’s quite hard to predict what we will use it for because even the creators of GPT-4 do not know what it can do,” says Bentley.
“The best prediction I can give is that this is the start of a new exponential. We had the dot com boom (and bust). We had the mobile app boom. Now we will have the large language model (LLM) boom.”
This is a boom that is already in full swing. Companies like Duolingo and Khan Academy have started to include ChatGPT into their apps to offer more personalised experiences.
Equally, payment software companies Stripe and Salesforce are using GPT-4 to improve their software, and even the Government of Iceland has opted into GPT-4, using it to help preserve the Icelandic language.
As Bentley pointed out we are very much in the middle of a large language model boom, but this is happening alongside a boom in artificial intelligence in general. There are ways to create images from scratch using AI, software to make your own 3D models, and even companies working on recreating your voice from a few soundbites.
Right now, OpenAI seems to be ahead of the curve, developing GPT-4 into a form of artificial intelligence that can revolutionise industries from web design to the legal profession, writing and comedy and even tasks like customer service, automation and learning a new language.
About our expert, Prof Peter Bentley
Peter is a computer scientist and author who is based at University College London. He is the author of books including 10 Short Lessons in Artificial Intelligence and Robotics and Digital Biology.
Read more about AI:
