Don’t worry, we weren’t forced to write that headline. Over the past few months, you might have witnessed murmurings about DALL-E or ChatGPT online. Not much going on, right? Just an obscure acronym of letters? Wrong. There’s an awful lot going on behind these acronyms – AI.
Now, when we talk AI, we’re not talking about Skynet or humans turning themselves into USB sticks. Put simply, AI (or artificial intelligence) is replicated intelligence in machines. Now, there’s some debate as to how artificial this intelligence can become. Should a sentient AI come about (more like Skynet), things could start to get… interesting. But we’re quite the way off from there – more at the beginning. But things are starting to get a little more exciting.
One of these tools, DALL-E, uses artificial intelligence to take a prompt from a
human person and create an image from it. But this tool doesn’t just have to understand what you’re saying, what this means, and what it looks like – that would be far too easy. DALL-E actually conjures up the image all by itself. Now, it takes inspiration from images on the web, all part of its training model. You can see this in some painting-esque images, where remnants of signatures can be seen. But the image it creates is original; you won’t find it anywhere else.
Another of these tools, ChatGPT, is more of a conversation-friendly implementation of AI. This language model interacts through conversation: answering questions, writing code, admitting its mistakes and rejecting inappropriate requests. People are using the tool for all sorts, from writing essays (though OpenAI will soon stop this with an embedded signature) to answering questions. It’s almost like you’re having a conversation. The tool has to understand what you’re saying, and then generate a response on the spot from its model of knowledge.
Now, there are plenty more AI tools out there – not just from OpenAI. Google is working on LaMDA for conversational AI, alongside all of the image and mapping stuff it does. And there are plenty more up-and-comers out there. All of these goings-ons signal something very exciting coming in the world of AI. But why should you care?
AI tools like ChatGPT are rather exciting because they have the potential to make our lives easier and more efficient. For example, ChatGPT can help with tasks like answering questions or generating text, which can save people time and effort. It’s almost like having your own personal assistant, but without having to pay anyone! Responses can be quicker and more accurate than those you’d get from people. Plus, it’s frankly pretty cool to see how advanced technology like this can be.
Or so says ChatGPT. See that above paragraph written in italics? That’s all ChatGPT. Pretty nifty, right? And it’s right! Whether it’s image generation or sophisticated chatbots, AI makes things an awful lot quicker, and requires less human interaction. The worlds of art and everyday tools are set to change in big ways over the coming years.
But it’s not entirely rosy… yet. These tools are currently in beta, as public previews for research purposes. As it stands, ChatGPT only has knowledge up to the end of 2021 – as that’s as far as its training model goes. DALL-E doesn’t quite get things right all the time, or at least not as we imagined them, as it can’t think independently. There are questions around whether AI-generated art is real or not, since a human didn’t make it. And what happens to consultants or assistants when ChatGPT can search the web and has information accurate to the second?
As you can see, there are still a lot of unanswered questions. But isn’t it exciting! Imagine a search engine with the power of ChatGPT behind it, giving more conversational answers. Unfortunately, that would probably be Bing, due to Microsoft‘s involvement in OpenAI – perhaps not the best example. Or imagine a smart assistant that’s actually… useful, and communicates like a normal person.
These things are no longer fictional, they’re in the pipeline. It’s time to start getting excited about AI – big things are coming. And we promise AI didn’t make us write this. Or is that what ChatGPT would say?