You, like most people will most definitely agree with the notion of A.I. ‘changing the world’, but how large is this impact really? How does OpenAI’s latest Chat Generative Pretrained Transformer (GPT) 4.0 play a role in this development?
You have most definitely heard of OpenAI’s revolutionary chatbot, mimicking human-like interactions with an inconceivable level of intelligence.
Most of you are likely familiar with ChatGPT 3.0, as are many schools, universities and corporations. This specific implementation of GPT, although revolutionary, still lacked the basic knowledge and common sense of humans. It was only able to score in the 10th percentile on the bar - a relatively low score for such an advanced piece of technology. Furthermore, it was quite rudimentary, being easily ‘convinced’ of concepts that simply are not correct or true. If you really wanted to, you could have easily convinced GPT 3.0 that 2 + 2 = 5. Some more malicious users found that they could convince chatGPT 3.0 to explain how to create bombs, drugs and poisons. Despite the world-changing applications of GPT 3.0, it was still littered with limitations, false logic, and a general lack of knowledge.
However, with the launch of ChatGPT 4.0, most, if not all of ChatGPT 3.0’s limitations have been entirely eliminated. The new rendition was seen consistently achieving in the 90th percentile on the bar exam. On top of that, GPT 4.0 has implemented a new multimodal feature, which will allow it to receive and interpret not only prompts, but images as well . For example, if you showed GPT 4.0 a picture of a basketball tied to a tree branch and asked, “What will happen if I cut the rope?”, GPT 4.0 will promptly reply, “The basketball will fall and bounce on the ground.” This display of artificial intelligence is none like we’ve ever seen before, having logic and problem solving far beyond any other system ever perceived. If you inserted an image of common ingredients like eggs, milk, flour, sugar etc, it would not only be able to identify all ingredients photographed, but could then generate a list of recipes that could be created from them. Though these tasks seem menial and simplistic, they in reality reveal a high level of cognitive thinking and reasoning. No other A.I. to date has shown the level of intelligence and competence that GPT 4.0 has.
Below are a list of things that the new ChatGPT 4.0 is capable of:
- Debug and write code in ALL coding languages
- Explain almost any concept, no matter how difficult or complex, simply enough that a 5 year old could understand it
- Extracting, summarising, and condensing any data given to it into any format required
- Creating entire Lawsuits
- Taking a hand drawn sketch and turning it into a functional website
- Generating new jokes and explaining why they are funny
- Saying “I love you” (if you work hard enough)
The figure above represents the amount of data that GPT 3.0 was created on compared to GPT 4.0’ data pool. When ChatGPT 3.0 was first released, the general consensus of the populus was that it was interesting, groundbreaking and all-in-all, incredible. However, one response piqued my interest far more than any other, “Terrifying”. These A.I. technologies have the capacity to do incredible things, but we must also ask ourselves, what will be the future implications of ChatGPT? What stability will jobs have in the future if a robot can write flawless code, create entire lawsuits and complete jobs that would otherwise take days for a normal person in a matter of minutes? What control will OpenAI’s ChatGPT have on the world?