Is OpenAI's Chat GPT really a threat to the world as we know it?

This article is aimed at discussing the most controversial invention of 2023 #chatgpt by #openai . There are numerous articles, Instagram spoof reels, WhatsApp hoax messages etc. authored mostly by unreliable sources. As someone having worked on their master’s dissertation in the field of #GenerativePretrainedTransformers (GPT), it (literally) pains me not to speak about this topic! So here we go!!

First of, what is Chat GPT?

To begin with Chat GPT is a conversational chatbot, like the ones we have on the flight booking website. Okay that's quite basic right? So, it makes us wonder what the hype is all about. Well, here's the secret of the soup, an American AI research lab-Open AI created an advanced version of this conversational bot, which basically employs Generative Pre-trained Transforms to estimate the probability of occurrence of a particular (token) word in a sentence and generates intuitive responses based on that probability.

Now what are Generative Pre-trained Transformers?

Well, simply put they are well crafted bunch of neural networks (building blocks of a deep learning model) which are trained using an exhaustive #trainingdata ( either in text or images). When training occurs in a #neuralnetwork model, they learn some patterns (or retain weights as we put it technically). These learnings (weights)are tuned and reused when the algorithm is loaded(activated), thus it makes it a pre-trained model. The quantity and quality of training data plays a key role in determining the response of the #chatbot. The more exhaustive, more relevant, more robust the training data is, the more apt is the response.

How does Chat GPT work?

So the underlying natural language processing model #gpt3 and #gpt4 are trained on vast data gathered from web scraping, open data sources(Wikipedia, webster dictionary etc.), social media and knowledge base(whitepaper publications, e-libraries etc) in a (controlled) regulated manner with human intervention, we term this as #supervisedlearning . We freeze these weights so that the model can be reused without reinventing the wheel. Next, we adjust (fine-tune) the model to make it more relevant to the real-world applications, using a feedback mechanism, where every question we ask is used as data to further perfect its response, this is termed as #reinforcementlearning. Using both these techniques the Chat GPT can generate an appropriate response, the relevance however is sometimes questionable!

What are its applications?

Apart from the infamous applications of it being able to help you cheat in your exams, write your term paper on your behalf, or modify job resumes. This advanced AI can be used for more serious purposes, such as #startups can gather end-user preferences, consumer behavior, customer satisfaction from many users in small amount of time, because the bot does all the work simultaneously in a jiffy. This data can be used by businesses to develop more targeted marketing strategies and improve their overall customer experience. It has been used in more serious application such as providing mental health advice and informing judge's ruling in a court case.

What are the threats concerned?

The above application raise the concern about the reliability and data security associated with the bot. There are plausible arguments that this chatbot has been released too soon, before enough of the general public is aware of the user manual or the interpretability of the bot's outcome. According to Professor Wooldridge, one concern is that the #technology “doesn’t have any conception of the truth. It gets things wrong in very plausible ways. It’s easy to believe and easy to be taken in by some answers it offers”. A real risk is this content is then used to spread misinformation. The chatbot is like molten steel which can be wielded in any manner as a butter-knife or sword based on the quality, integrity and non-discriminatory nature of the training data. It is important to spread awareness on the fact that the chatbot can be biased and unreliable based on the training data, so it is not superior to subject matter experts, #datascientists, #AIEngineers,doctors, lawyers or professional, it can only aid their daily activities to remove redundant tasks.

The Golden Question of the Era: Will it replace human job force?

In the words of the Founder of #infosys , #NarayanaMurthy, "ChatGPT can’t beat the human mind". His genius is on point! Chat GPT like other #aiproducts needs the #humanintelligence to back in terms of regulating its content for #humancentereddesign and #humancentric #decisionmaking . The biggest boon of human kind is our #empathy and #emotionalintelligence , which at the end of the day roots from the human consciousness. This is the core of any type of #businessstrategy or #governmentpolicymaking which requires the consideration of the human element, external and internal factor of an organization of individual and (most importantly) common sense! The chatbot might replace redundant clerical jobs in many sectors such as customer service, copy writing, legal ,healthcare etc. however, a professional or expert is still required to assess the outcome and make #appropriateapplications of this outcome.

In conclusion, the human genius is far more superior to the #generativeai in context. However, it is necessary for us to apply our common sense and exercise the most complex system ever build - the human mind. The irreplaceable asset we human possess makes us not only the most intelligent being on the planet but also, the most feared of them all, which is the #naturalintelligence of the human nervous system!