Nov 15, 2020
AI can write your code, are you scared codder?
What if I told you, a day will come in the future where there is a super AI that can write Complex Machine Learning Algorithms by itself, write code for programing challenges to building a full-scale functional website by simply telling it what we need, What if there is an AI that can not only write articles but can imitate people by how they speak by simply reading the books they wrote.
THE SUPER AI THAT I TOLD YOU THAT WILL BE REAL IN THE FUTURE
Well, I Lied. It’s already here
Let me introduce you to GPT-3, It is made by a company called openAI which is cofounded by Elon Musk. It is one of the most advanced AI that our Human rays have not only built but achieved. It's one of the greatest milestones in the last 6 million years the humans have achieved.
GPT-3 is the only machine learning algorithm that was trained with this huge dataset with Billions of Parameters.
GPT-3 is an allrounder
GPT-3 is an Author, Blogger, Codder, Comedian, Quizmaster, and a lot more. It is also very good at answering both the simple and complex analogies where most of the AI robots fall short on. if we ask GPT-3 about a simple analogy like alphabetical order or things that make sense to us but doest make sense in a logical or arithmetic way which in return makes it hard for robots to understand it. There is a scenario that happened where it was given a complex set of data like how bitcoin works and asked to explain it to a person with no prior economic or technical knowledge and GPT-3 had really done a fantastic job in explaining how bitcoin works to a person that doesn't know much about economics or software, It is also good at explaining the compute program or code and how it works and also give suggestion to the programmers while writing the code. It can also simply generate the face by simply giving a beef description to GPT-3 and it will generate a fake face using deep fakes. These kinds of AI are really a cornerstone and milestone to our human civilization itself. The multi-talents of GPT-3 is really amazing and watching them in real-time is like watching an AI in a sci-fi movie. Wave at this blog if you agree
GPT-3 is one of the kind. one time a Twitter user had a conversation with Einstein(GPT-3 is imitating Einstein). I was surprised when I first had a glimpse at the conversation chart between them because it's not like an imitation bot and a human conversation. It's like Einstein himself has visited the Future to have this conversation. At first, the conversation is pretty sobber but it becomes really interesting when they go into technical details like the discussion of the theory of relativity and the relation between space and time. Because those adaptations made by the GPT-3 are so much close to the actions of Einstein
Ethical Concerns
A power AI like GPT-3 can not only imitate people but can also imitate dead people as you can see from the above example of a conversation between Einstein and a Twitter User. This raises all types of Moral and Ethical concerns like ‘is imitating dead people using AI’ a good idea.
There is a Netflix Hit Series called Black Mirror. it's a series that takes horrible scenarios that may happen due to the development of technology and shows the dark side of technological development. in that series, there is an episode where the main protagonist in the Episod loses her husband and is not able to handle the death of her partner well, then she comes across the Software which talks like the person you want to speak with(by reading their messages, chat history and emails the software imitates a person), She slowly gets attracted to that software and spend most of her time by charting with that robot(robot imitating her husband) and after some days she could not stop talking to him and slowly she gets into the loop of emotions and no matter how hard she tries she could not pull herself out of that emotional pit and slowly she turns Insane...This may be a hypothetical scenario but still, the matter of fact is still using this kind of software that may end up in these types of possibilities in mind-boggling and the Truth is these kinds of Softwares Already exist in our world.
GPT-3 is so powerful that it can imitate the writing of the author and write poems by simply reading some poems written by the author and it can not only create Articles but college students can actually abuse this and make the AI(GPT-3) complete their assignments, Home works, labs, and whatnot. As a matter of fact, this entire blog is written by GPT-3…Just Kidding…or am I
Can GPT-3 automate software Jobs?
GPT-3 can generate code in any software language and can make an app or website by simply telling it what kind of layout you need by describing it, is a real thing that software engineers need to worry about, it can also explain what is the specific function of that line of code at that particular place and it can also give suggestions to the developers when developers are writing the code. but still, even GPT-3 is also reading the code from previous developers through the data from its dataset and generating the code from those patterns. So, No software developers don't need to worry about the Automation of the software development field…But still it doesn't pain to have an eye on it.
How close are we to achieve a super AI?
There is a test called Turing Test to answer the question ‘can machine think?’. the test is conducted in such a way where there are judges, participants, and a machine(in which this turning test is performed) are isolated the judges don't know whom they are interviewing(either participants or the machine) if the judges are not able to tell who are participants and what is the machine by the replies they gave then the test is successful. And the turning test for GPT-3 is walk in a park because it can already imitate people and do way beyond that that the Turing test expects from the machine, but still, there is a huge limitation for all the AI robots we build including GPT-3.
Limitations
Even if all these advanced AI’s can pass the turning test and can write articles or do projects or complete assignments like humans at the end of the day all these Robots are understanding the patterns in the way humans do it and simply replicating it from the data of huge datasets that we provide to it. All these advanced AI’s cannot think like human’s, they may write articles like humans by simply understanding the patterns by which human writes them but can are not able to understand what that article and the info behind it really means, they may understand unstructured data like images, languages, Audio conversation by understanding the patterns but are not able to get the significance of that unstructured data as they cannot think like humans. Even the self Driving car of today can only understand the patterns of work that must be done through reinforced learning but are not able to understand the reasoning of the work it does.
In fact, as the GPT-3 doesn’t actually understand the significance of the data from the dataset and only understand the pattern from it it will make some non-sensical mistakes
But the million-dollar question at the end of the day is “is our fear over AI Justified ?” Well, the answer is it little more complex. There is no Red pill or Blue pill here, your fear is justified in some cases and not justified in other cases. Yes, we need to be scared of AI and its rapid development and Automatisation of most industries due to Automatisation as more and more jobs are becoming Automatised like Driving, Auto Mobile manufacturing, Agriculture, etc… and also No, as the current AI in our current technology can only understand the data by seeing the patterns but cannot think by itself.
References:
https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3
