By Olivia Gibson
On 13th May 2024, the leading tech company OpenAI released the latest model of ChatGPT, version 4o. In the past couple years, artificial intelligence (AI) chatbots like ChatGPT have really taken centre stage in the tech industry and within wider society. The capabilities and potential of AI are continually assessed and are often a source of controversy.
Society currently faces issues of changing plagiarism laws and having to redefine intellectual property. These issues particularly concern students and the education sector as a whole. Therefore, with the release of ChatGPT-4o, it is more prudent than ever before that students are aware of what ChatGPT actually is, what ChatGPT can really do, and what do these advancements mean for students in the long term?
First released by OpenAI in late 2022, ChatGPT is a form of AI that uses natural language processing to provide a human-like assistant. Simply, natural language processing is a computational method of analysing and ‘understanding’ human speech and text. This allows ChatGPT to respond to our text inputs no matter how formal or informal they may be.
The ‘GPT’ in ChatGPT stands for ‘Generative Pre-Trained Transformer’ which is a type of Large Language Model (LLM). An LLM is an artificial neural network that mimics the human brain by being able to respond to situations based on previous, related experiences. In the case of ChatGPT, the responses to requests are generated based on the most positive feedback of other, similar requests.
The new update allows image and audio inputs
The ‘Pre-Trained’ aspect comes from its training on a large range of text taken from the Internet – prior to release – so that grammar, reasoning abilities and general, key facts can all be learnt. This is why it was originally specified that ChatGPT only contained real-life facts up until a certain date (the date at which the pre-training ended).
In essence, ChatGPT is trained in the art of human language and facts about our world and cultures, so that when an input is received, it can process this and generate a response based on what it deems is most appropriate.
So, what makes the newest version, ChatGPT-4o, so special?
The biggest addition to ChatGPT-4o is the ability to have different media inputs. Rather than just text-based inputs in ChatGPT-4, you can now give ChatGPT-4o images or audio. This increases the complexity of what ChatGPT can be asked, for example, being able to screenshot mathematical formulae rather than struggling with formatting.
Additionally, the conversational ability of ChatGPT-4o is much improved; it can understand different tones, multiple speakers, and background noises. ChatGPT-4o can also respond with interactive audio and visual aids making its explanations much more accessible and clearer.
On a technical level, ChatGPT-4o has a much faster response time that is very similar to human-conversation levels. Overall, we can see that ChatGPT-4o is getting increasingly closer to truly mimicking human interactions.
Of course, ChatGPT-4o is not perfect and certainly not without limitations. Indeed, asking ChatGPT anything rather niche still results in dubious answers conveyed with such confidence that it becomes quite difficult to spot where faults occur. Additionally, the image and audio processing is new and will take time to perfect.
However, what we should take away from the release of ChatGPT-4o is that even with such intelligent software, the advancements in generative AI chatbots are still developing and there is much to be uncovered.
The overwhelming reception from the industry after the release of ChatGPT-4o has been very positive. Certainly, a potential partnership with tech giants, Apple, is indicative of how in-demand ChatGPT is and how well it has been received.
However, with every leap forward in Artificial Intelligence, there is always the tail of a fundamentally unanswered question: how is this going to integrate into society?
The current rule for students is, generally, ChatGPT is forbidden for aiding assignments, although no such rules exist for educators in their preparation of lessons or lectures or exams. With every update of ChatGPT, the new features are deemed almost useless as students cannot utilise them for their primary source of work. The use of ChatGPT for an assignment is made to feel almost ‘criminal’.
Despite this, in recent years, education has seen a rapid increase in students using generative AI to aid their studies. An early 2023 study by Forbes showed that 89% of students have used ChatGPT to help with homework. A figure that has potentially risen nearly a year and a half on.
The conflict between students being unable to resist using such a life-changing piece of technology to aid their studies (somewhat akin to the use of a calculator) and the current rules forbidding the use of ChatGPT is only growing more strenuous.
The release of ChatGPT-4o calls for a serious discussion at the highest level in academic institutions and within the Department for Education about how generative AI can be used alongside education, rather than against it, as its rapid improvements means that its use and popularity amongst students is inevitable.
Image: Jernej Furman via Wikimedia Commons
