Many wonderful technology have vanished more quickly than we anticipated. It is quite challenging to foresee changes in the digital world. Many of us never would have imagined the internet being the phenomenon that it is today. Newer skyscrapers built with the most recent disruptive technology are rising faster than ever in the ever-deepening graveyard of antiquated technology.
The ChatGPT has been dazzled the web world for the past few months with its astounding ability to write text that reads like human speech, contextualise ideas, and adapt to conversational manner of communication so well. For the past two decades, Google search has been the most popular method for quickly satiating curiosity. But this innovative, free service powered by artificial intelligence appears to be upending how regular people have sought for information.
When we enter search terms into Google, a variety of third-party links connecting to websites are given as results. More recently, though, it has begun to emphasise portions of websites that match the search terms. The user profited from this trade since it gave them access to quick information, and later Google employed AI models to anticipate and contextualise the results based on demographics. But it didn’t seem like a dialogue, which was possibly due to the lack of time.
Google emerged at a time and place when the digital world was largely viewed as a set of tools for boosting productivity and was not yet widely embraced in everyday life. As the battle between the virtual and the actual world was about to end, ChatGPT entered the stage. People now view digital technologies as basic necessity and wish to make them more resemble human beings. a suitable environment for conversational AI models.
| Google Search | ChatGPT |
| Produces search result based on keywords | Based on conversational model, feel like human interaction |
| The result includes links, excerpts from webpages, Google Profile and videos | The result mostly is well formatted text with links if required |
| Is updated with the current affairs | Is only limited to offline data obtained through LLM (large language model) till 2020 |
GPT(Generative Pre-trained Transformer)-2 to GPT-3
LeNet, Yann LeCun’s ground-breaking neural network, had 60,000 parameters in 1998. After 20 years, the initial GPT version from OpenAI contained 110 million parameters. GPT-3, which is two years old, has 175 billion, whereas GPT-2 has 1.75 billion.
29 months have passed since OpenAI released GPT-2, a large-language model that demonstrated the effectiveness of transformer-based neural networks. Experts were astounded by the quality of GPT-2’s natural text output. GPT-3, which was announced in 2020 as a replacement, was a bigger, more complex model that generated even more powerful results when used with not only texts but also sounds, images, and movies.
Emergence of ChatGPT-4
Though there is no formal confirmation of the debut or beta testing of OpenAI’s ChatGPT-4, which is expected for 2023, anticipation is growing around it. Generative Pre-trained Transformer 4 is known as GPT-4. In essence, it’s an AI programme that can produce text that sounds like human speech.
According to some reports, ChatGPT-4 could feature 1 trillion parameters, up from the current ChatGPT-3’s 175 billion. Like the present ChatGPT-3, it will also be capable of text replying, content creation, language translation, and text summarizing.
Conclusion
The ever demanding user population will always seek out newer and more convenient ways to produce and transport data. As long as the technique efficiently conveys the message, the method loses importance. However, the primary internet gateway “Google,” which also became synonymous to World Wide Web, has likely served its purpose and it is now time for a novel method of interacting with the rapidly developing internet. Despite the fact that the conversational models are promising, it is still too early to predict that they will replace Google Search.
