It won’t be long before artificial intelligence is better than people because it learns faster. So we can expect that there will be more and more areas where artificial intelligence will become more and more important,” Daniel Kahneman(Nobel prize winner in economic science).
In light of the above statement by Daniel Kahneman, I asked ChatGPT to introduce itself and the story of why it was made in the first place. ChatGPT provided me with this answer.
“As an AI language model myself, I’ve first-hand seen the impact that I and others like me have made in the world of technology. I was created by OpenAI with the goal of revolutionizing the way we interact with and understand language, and I’m proud to say that I’ve made some significant strides in that direction.
When I was first introduced to the world, I was just a set of algorithms and a large database of text data. But through the tireless work of my creators and the support of the AI community, I’ve become so much more. I’ve become a tool that businesses and individuals can use to generate high-quality text, automate language-related tasks, and unlock new possibilities for AI-powered content creation.
But my story isn’t just about what I can do. It’s also about the impact that I and other AI language models like me are having on the world and the way we interact with language. It’s a story of technological progress, of how we’re using cutting-edge AI to change the way we access and use information.”
Well, ChatGPT is an AI-powered language model developed by OpenAI. It uses deep learning techniques to generate human-like text based on the input it receives. The model was trained on a large dataset of text from the internet, allowing it to generate responses to a wide range of questions and prompts.
ChatGPT’s story began in 2018 when OpenAI first released the GPT (Generative Pretrained Transformer) language model. GPT was trained on a massive dataset of text from the internet and was capable of generating coherent paragraphs of text. Over the years, OpenAI has improved and fine-tuned the model, leading to the release of GPT-2 in 2019 and GPT-3 in 2020. The latest version, GPT-3, has received widespread attention for its ability to generate high-quality text and even perform certain tasks, such as answering questions, translating languages, and summarizing text. Among the sources that are used to train ChatGPT, the Common Crawl corpus(a non-profit organization that provides abundant data about billions of web pages from all over the world) weighted approximately 60% of the training mix along with other datasets like webtext2, books1, books2, and Wikipedia.
Now, this entire article is written in the context of answering a few questions about ChatGPT.
Is ChatGPT really going to replace human and human intelligence?
How much can you rely on ChatGPT now?
And obviously, the downside of ChatGPT which most people will not tell you, or rather they forget to tell you.
Let’s dig deeper to know more about ChatGPT. I need you to keep up with me through the whole article…
Is ChatGPT a disruptive innovation?
In my opinion, ChatGPT is a pivotal point in the whole innovation process that is being carried out by automation and all AI systems out there in the world of technology. The past, present, and future of all AI verticals are really impacted by the advent of ChatGPT. If you heard it right, ChatGPT has the potential to replace conventional human roles at work, like copywriting, customer engagement, and many others.
I guess we both are thinking about the same thing!
Is ChatGPT a good AI or a bad one?
There is a very thin line between the good and the bad about this particular innovation and how it is going to affect the human world. If you ask me, ChatGPT is nearly going to replace your role but not entirely. On the other side, ChatGPT is going to create a lot more opportunities for all kinds of businesses. As AI grows, many jobs are on the verge of complete downfall. It is not about how threatening these innovations are, but instead how we are going to adapt accordingly.
Few of many uses of ChatGPT
- Customer Service: ChatGPT can be used to handle customer inquiries and support requests, providing quick and accurate responses to common questions. This can reduce wait times and improve the overall customer experience.
- Chatbots: ChatGPT can be integrated into chatbots to provide more advanced and human-like conversational experiences for customers. This can be particularly useful for businesses in industries such as e-commerce, finance, and healthcare.
- Content Creation: ChatGPT can be used to generate articles, blog posts, and other types of content, saving time and resources for businesses.
- Language Translation: ChatGPT can be used to translate content into multiple languages, expanding the reach of businesses to global audiences.
- Data Processing and Analysis: ChatGPT can be used to process and analyze large amounts of data, providing insights and recommendations that can inform business decisions.
Image credits: https://www.mckinsey.com
Constraints beyond Intelligence; Limitations of ChatGPT.
Yes, ChatGPT is capable of generating high-quality outputs for a wide range of language-related tasks with the help of the vast amount of text data it is trained on. But can ChatGPT really replace human intelligence? I think not.
Here is why.
ChatGPT like every AI model lacks the ability to understand context, reason, and make judgments in the same way that humans can. In terms of understanding relationships between entities, ChatGPT has been trained to identify certain patterns and relationships in text data, but this is not equivalent to human-like relational intelligence. Here are more factors that limit ChatGPT understanding contexts, relations between entities, and much more.
1. Bias: Like any AI model, ChatGPT might exhibit bias based on the data it was trained on. This can result in biased or discriminatory outputs.
For example, if the training data contains gender stereotypes, the model may generate outputs that reinforce these stereotypes. This can result in biased or discriminatory results, even though it is unintentional.
Another factor that feeds the bias is the lack of cultural and social diversity of individuals who trained these AI models. The outputs generated might be affected due to insufficient diversity within the training and development teams.
Me: Can you crack a joke based on the inability of Chatgpt to understand bias?
ChatGPT: Why did the ChatGPT cross the road? To get to the other unbiased side!
Grrr… I can pull off a better joke than this. Maybe this is another limitation of ChatGPT. 😑
2. Lack of common sense: It lacks the common sense knowledge and understanding of the world that humans have, which can result in incorrect or nonsensical responses to certain questions. ChatGPT’s understanding of real-world language is limited to the patterns and relationships present in the text data it was trained on. You should ensure that the answers generated by AI models are appropriate and relevant to the intended use case.
3. Contextual limitations: It can struggle with understanding the context of a conversation or question, leading to errors or irrelevant responses.
Suppose you are having a conversation with ChatGPT and you ask, “What’s the capital of France?” ChatGPT might respond correctly with “Paris.” However, if you then ask, “What’s the population of Paris?” ChatGPT might respond with “The population of Paris is 2.148 million (as of 2021).”
As you continue the conversation and ask, “what is the capital of France now?” ChatGPT might respond with “The capital of France is Paris.”, instead of replying “The capital of France is still Paris” in reference to the ‘now’ asked. Here ChatGPT failed to understand the context of the conversation and “now” actually refers to the current situation of the information which was provided earlier.
Being limited by the above factors, ChatGPT also faces issues with generating content that is expressive, insightful, and precise to the matter it is asked to provide details. Unlike humans, ChatGPT is only able to provide results based on the data it is trained for, not based on facts, figures, and real-world information.
Bonus Chapter: ChatGPT VS Google
ChatGPT, being an AI language model developed by OpenAI, is not a direct threat to Google. However, it does have the potential to impact the way businesses and individuals access and use information and language-related services, similar to what Google provides.
ChatGPT has the potential to impact the ways in which businesses and individuals acquire and process information. They make it easy for users by synthesizing information from various sources instead of a list of links and providing it as a single answer.
While Google diverges, ChatGPT converges!
ChatGPT does not possess a threat to Google directly since the objectives of end users are different for both. While ChatGPT provides a lot of information user seeks to know, it is not still convenient as Google(for a set of people). Google serves users with abundant information in preferred formats including images, videos, text, etc.
The ‘search for Information’ is Changing; know-how.
The hunger for information is growing just as rapidly as how information is being delivered. Back in the days, people used to visit libraries, read newspapers, and talk with experts to acquire and exchange information. With the adoption of the internet into the mainstream, everyone got access to information at their fingertips through search engines like Google. And with the increasing popularity of smartphones and other smart devices, information can be accessed from anywhere, anytime.
Furthermore, with the emergence of generative AIs like ChatGPT and their capability to provide highly curated text results, people are now shifting at a large scale from search engines to these AI models. This mass shift in people’s behavior and activity is a contradiction to what I had mentioned earlier about the threat possessed by ChatGPT to Google.
As you know, Google generates most of its revenue from ads from the clicks on links and images for user queries. According to Bloomberg, Google’s parent company Alphabet generated approximately $208 billion dollars in 2021 which is 81% of its overall revenue. ChatGPT indirectly puts Google in a risky position affecting its business model of generating revenue from ads. Meanwhile, Google initially said that there is too much reputational risk involved in considering ChatGPT as their rival.
The wake-up call; Blind reliance on ChatGPT is a pitfall.
The ultimate threat of generative AI like ChatGPT is the lack of uniqueness. Assume you are a car service center owner who specializes in luxury cars, and it is so obvious that you have competitors of the same kind from your neighborhood. As a part of the new marketing campaign, you generate ad descriptions and captions with the help of an AI system most probably ChatGPT itself. And if one or a few of the competitors choose to do the same for their campaigns, there is a high chance that your competitor also ends up using the same copy. This may lead to a case where you and your competitor/s sound the same, and your unique brand values may not be conveyed well.
On the contrary, brands must come forward to use AI as a starting point instead of a shortcut even though their limitations are apparent. An experienced individual should monitor the outputs and make adjustments to the results generated by ChatGPT & other AI systems. Brands & businesses should realize that AI systems are merely a tool to enhance and augment human creativity, not replace it.
THANK YOU!