Motivation
AI is rapidly becoming prominent in society today. This begs the question are many jobs and roles going to become redundant soon?
ChatGPT and the GPT Language Model
A new language model that has been trending recently is the Generative Pre-trained Transformer (or GPT for short). I asked the AI model what it was and this is what it came up with:
“I am not a chatbot or a product called ChatGPT. I am a large language model called GPT (short for “Generative Pre-trained Transformer”) that was developed by OpenAI. GPT is a type of artificial intelligence that is able to generate human-like text based on the prompts given to it. It can be used to perform a variety of tasks, including generating responses to user input, summarizing text, and translating languages. GPT is trained on a large dataset of text and uses machine learning techniques to learn the patterns and structures of human language. It can then use this knowledge to generate text that is coherent and resembles natural human language.”
How it was Trained:
ChatGPT is fine-tuned from a model in the GPT-3.5 series, which finished its training in early 2022. This model was trained using Reinforcement Learning from Human Feedback (RLHF). This meant that OpenAI had human AI trainers that provided conversations where they would play both sides. A reward model is required for the AI to correctly train it. This was done by taking conversations the AI trainers had randomly select a AI written message then the human trainers ranked this message among other alternative messages for the same prompt. This was done iteratively to continue to fine-tune the model.
Chat GPT at Work:
ChatGPT writes me a blog post
To show the power of this AI I put it to the test and asked it to write me a blog post.
The prompt:
“write a blog post on coders jobs being taken by ai and the prominence of ai”
The result:
“Artificial intelligence (Al) is increasingly being used in a variety of industries, including the
field of computer programming. As Al technologies continue to advance, there is a growing
concern that Al may eventually replace human coders and take over programming jobs.
There are several ways in which Al is being used in the field of programming. One common
application is code completion, in which Al algorithms can predict and suggest code
snippets based on a programmer’s input. Al can also be used to automate routine tasks,
such as debugging and testing code, freeing up programmers to focus on more complex
tasks.
However, there are also concerns that Al may eventually become advanced enough to
replace human programmers altogether. Some experts believe that Al will be able to write
and debug code more efficiently than humans, potentially leading to the displacement of
human workers.
While it is difficult to predict exactly how Al will impact the field of programming in the
future, it is clear that it will play a significant role. It is important for coders to stay up to date
with the latest Al technologies and to be proactive in developing their skills in order to
remain competitive in the job market.
At the same time, it is important for organizations to consider the potential impacts of Al on
their workforce and to take steps to ensure that the benefits of Al are shared widely and that
any negative impacts are minimized. As Al continues to become more prominent, it will be
important for everyone to carefully consider the potential impacts and to work together to
ensure that the development and deployment of Al aligns with our values and goals as a
society.”
AI Codes from plain English:
The ability of GPT-3 is shown further in this tweet for Sharif Shameem showing how you can generate JSX code purely from plain English prompts. Jobs like coding may take a different turn in the future. Where the coder is required to think and find solutions rather than spend the bulk of their time working out how to code said solution.
This is mind blowing.
— Sharif Shameem (@sharifshameem) July 13, 2020
With GPT-3, I built a layout generator where you just describe any layout you want, and it generates the JSX code for you.
W H A T pic.twitter.com/w8JkrZO4lk
My thoughts:
As you can see the possibilities of this AI model seem endless, all that is needed is time for it to train and it will only get better.
I think that it is great advancement to have such models that can do these things. For example an AI like this could replace a search engine like google making it much easier to receive digestible information saving the user the time in researching. For example, it was able to explain complex topics, like “Potential Flow” in a digestible way. However, the AI may include biased content so the human may not get correct answers to subjective questions at least for now. Also, problems arise in the academic fields where plagiarism or academically dishonest work isn’t acceptable. For example in coding exams the use of AI makes the test redundant as the aim of the test is to test the ability of the applicant rather than the AI. However, as said before the rise of these AIs may change the role of programmers and hence the exams required to test their ability. Overall, I feel the development of AI will help us become more efficient however there are still lots of fields where the use of AI will not be acceptable but being able to stop it will be hard to achieve.
