The new realm of coding with GPT-3
It had been a while since my last blog and through this time, I have gotten sometime to read over the technology of tomorrow. Most of us following my blog are find of some of the major coding languages, such as Python and Java, however, there is something new around the table that will change the game in this aspect. That thing is the new technology GPT-3(Generative Pre-trained Transformer 3) by OpenAI.
What makes this language different from the rest?
When coding, you have to write out the code lines and have to think about what to write after each function. However, GPT-3 does not require that. Picture this: you are about to write a line of code that will bring out a value from the for-loop. When you write print, the code will automatically give you methods to fill out the rest of the line. One of the options will allow you print out the specific value and so, you can autocomplete the line. Although it may not seem as that big of deal when reading about it, GPT combines the functionality of coding and the efficiency of auto-fill that we commonly see on Google Search.
Another feature of GPT is one that will surprise many. In GPT, you have the ability to translate regular english into lines of code. For example, if I say I want a button that looks like a watermelon, I will get the lines of code in HTML that will make a visual button that looks like a watermelon. Like what! That may sound Science Fiction to many of you guys but it will be possible one GPT is released to public. Right now, it is a sign-up based on a first-come-first-serve basis.
In addition, for those who do not know, the great Elon Musk is behind this new transformation in coding. He aims to make coding much more simpler to many and aims to power this new generation of Software Development with AI. But, without any proper testing available, questions are being asked.
Will jobs in IT be lost because of this new change?
With the concern of the code automatically creating the software, people are worried that programmers will loose their jobs as robots take over the coding world. However, GPT-3 is far from perfect. The auto-complete code that it provides still experiences errors here and there. It requires skillful engineers to fix the problems and errors that are created by GPT. With this, more jobs will be created where Programmers will analyze the code and will fix the bugs present in the code. In addition, the simplicity of GPT-3, when it is up and running, will allow employees to focus less on the coding aspect of their job and more on the creativity and innovating in countless ways. All in all, OpenAI has created a good stepping stone, especially in the IT World.
Where else is GPT-3 Applicable?
I have only went in-depth on the applications it has on the technology business, but it has implication in the fields of graphic design, languages, and other aspects where AI and ML is applicable. Here is an example of the functionality of GPT-3:
In this demo, we can see that the user types in what he wants his application layout to be in simple english and with the power of AI, the software was able to decipher a solution that similar to the UI in Instagram(the software used to create the app is Figma which is something I used in my internship as well).
Another demo of the software is in this article, where the user asked GPT-3 to write an essay convincing people that robots come in peace(Credits to The Guardian and GPT-3 for writing the essay):
How does GPT-3 work?
We have discussed the various application of GPT-3 but we have not discussed the actually operations and processing power of GPT-3 and how that makes all of this possible. In essence, GPT-3 is a language prediction model, meaning it takes an input by the user and translates/transforms it into to most useful piece for the user. The software has been pre-trained by OpenAI, where the company has expanded the huge amounts of compute resources necessary for GPT-3 to understand how languages work, their syntax/diction, and the structure. This analytical thinking in the technology allows the compute to decipher what the user is saying based on the specific grammar and syntax used in the sentence and accumulates data based on these phrases and sentences. This is a very complicated process as there are thousands of words, each with various different meanings but the computer continues to learn whenever something new is inputed into the system.
GPT-3 also uses a dynamic “weighing” process, using one of the largest artificial neural networks ever created. As it weights its output according to the input given by the user, the weights are held in its memory and uses the process query each(there can 175 billion weights held).
Where the future lies…
With technology rapidly evolving each day, it is time for the people to get ready for this change. GPT is one of the many ways technology is changing and is something we should our eyes on. With coding becoming even more easier and with the implementation of AI, more jobs will open up. The future of AI/ML is direction unknown. To participate and get a beta version of GPT-3, use this link to request and experiment with the latest and greatest technology.
