Edward Sutton – Head of Science and Tech
As the end of the academic years draws to a close, I wanted to reflect on one technology that has had a renaissance this year. That is AI – which has changed from feature buzzword to a completely new beast that has changed the way we approach the internet.
Start of the year, websites like Chat GPT seemed like a new buzzword that was floating around Silicone Valley. A new way to search and create from just a few words. The possibilities seemed endless – from simply asking a question to much more complex tasks like generating code. Now at the end of the year, the Large Language Models like Chat GPT, Bing and Google Bard are everywhere and available to us all at the click of a button. The power these new AI tools have are incredible and are only getting better. With the ability to create entire essays and articles from just a little prompt, AI has now raised new questions about the nature of work and the necessity of certain jobs in the future.
2023 is the year of AI but with so much advancement on the horizon – this is looking more like the decade of AI
One of the most significant developments came in 2021 when OpenAI released Chat GPT-3, an updated version of the language model that was 10 times larger than its predecessor. This allowed Chat GPT-3 to generate responses that were even more natural and human-like.
Another significant development came in 2022 when OpenAI introduced Chat GPT-3.5, an even more powerful version of the language model. Chat GPT-3.5 was developed using a new type of deep learning algorithm called generative adversarial networks (GANs). This allowed Chat GPT-3.5 to generate responses that were not only grammatically correct and natural-sounding but also more creative and imaginative.
Pandora’s Box of AI is now open and the scary possibilities for disinformation and impersonation is one of the new technology’s dark sides
The development of Chat GPT has not been without its challenges, however. One of the biggest challenges has been ensuring that the language model is not biased in its responses. Bias can occur when the AI algorithm is trained on data that is not representative of the entire population. To address this issue, OpenAI has implemented a number of measures to ensure that Chat GPT is as unbiased as possible.
Now, would it surprise you to know that the last two paragraphs of this article were written by Chat GPT? You might have realised but for some people they would not have blinked twice. The fact it can write such a convincing paragraph is slightly scary as I sit here at my desk trying to do the exact same thing but taking 3x longer. How tech companies are framing these AI models is a way to free us from the mundane and the drudgery of work and allowing us to enjoy more of our day. For people in the creative industry, however, it is an encroachment on the very soul of their profession.
Pandora’s Box of AI is now open and the scary possibilities for disinformation and impersonation is on of the new technology’s dark sides. You might have seen that viral track of AI Drake, but this is just the beginning of a new voice imitating AI. As its always improving with more data, this will only mean more natural sounding voices of not just singers but also politicians. The worry that bad actors might use these voice models to create fake speeches, alongside generated images, opens a whole new challenge to democracies all around the world.
This technology is undoubtedly exciting and a leap in technological development that we can all use for ourselves. However, let us hope that this time next year I am not writing about the constant issues created by AI – whether that be political or creative. 2023 is the year of AI but with so much advancement on the horizon – this is looking more like the decade of AI.