ChatGPT has just hit 1 Million users in less than a week of being launched, so what is ChatGPT? Over the weekend, you may have noticed some of your online friends sharing some computer-generated conversations and wondered what’s all that about.
ChatGPT is the latest AI project from OpenAI, the people that brought us DALL-E 2, the tool that allows you to generate images, just by describing the image you want.
Open AI has leveraged the learnings from GPT-3 and built an incredibly sophisticated language model that allows users to interact interacts in a conversational way.
The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.
ChatGPT is trained on an Azure AI supercomputing infrastructure and will get better over time, but even today, it’s capabilities are incredibly impressive. It’s not new for computers to be able to generate information based on a set of inputs, however, most of the chat systems you interact with, like support bots on websites, are a simple web of if, this, then conditional statements that return a pre-determined set of responses. Anyone who’s spent time interacting with these systems quickly understands this limitation in just a couple of minutes.
ChatGPT is vastly superior in its ability to understand sentence structure, and the meaning behind words, allowing its responses to be much more human-link and much more useful.
Take this example below, where the user asks ChatGPT to write programming code and the AI does it. This isn’t just some party trick, this could dramatically accelerate the development cycles for application and web developers.
When we think about the types of careers that computers and by extension Artificial Intelligence will disrupt, we like to imagine that creative employment will go largely unaffected. After seeing what people are using ChatGPT for, it’s likely that essays, books, movie scripts and even blog posts are all squarely in the crosshairs of AI.
On OpenAI’s page for ChatGPT, they do recognise things aren’t perfect and the system does have a number of shortcomings.
One example of this is this – The model is often excessively verbose and overuses certain phrases, such as restating that it’s a language model trained by OpenAI. These issues arise from biases in the training data (trainers prefer longer answers that look more comprehensive) and well-known over-optimization issues.
It’s important to understand this is just the start and AI is iterating and improving quickly. Exciting times are ahead.
If you want to try out ChatGPT, unfortunately, you may struggle as their servers are overwhelmed by the demand to try it out.
CEO of OpenAI, Sam Altman, made the announcement that ChatGPT had crossed the 1Million mark on Twitter and got a reply from Elon Musk who founded OpenAI, asking how much the average cost per chat costs. The answer is just a few cents per chat.
Let’s hope Musk has a few dollars left over after buying Twitter to help scale out this amazing technology for the world. Also, don’t be surprised to see some kind of Twitter integration.. aka.. ‘Create a GIF that has a monkey playing pong using his mind’.