If ChatGPT’s viral popularity is any indication, conversational search is the way of the future.
Have you ever had to go through several pages of search engine results to find the solution to a challenging question? Let’s say you want to know if your dog can follow a vegetarian diet. You may start your study by going to Google, putting “is a veg diet healthy for dogs,” into the search bar, and then you’ll need to sort through the plethora of results that are returned. By the time you discover the solution, you’ve spent much longer than you anticipated reading papers, reports, and their sources.
Finding the solution to a challenging topic might not be such a laborious, mind-numbing procedure in the not-too-distant future. According to reports, Microsoft is merging the AI technology that powers ChatGPT into its Bing search engine, a move that might revolutionise search as we now know it. According to AI experts, Bing may have the ability to deliver a search experience that is superior to Google and potentially challenge the search giant’s long-standing hegemony.
According to Anton Korinek, an AI researcher and professor of economics at the University of Virginia, “ChatGPT is the first new technology in more than a decade that may actually revolutionise search and that may, at least in principle, upend Google’s market domination.” The technology enables users to communicate with their computers in a way that is far more conversational and natural than traditional search, according to the company.
We don’t yet know exactly how Bing’s AI-driven search results will appear. For this report, Microsoft declined to comment. However, in terms of how a search engine delivers a result and how people engage with it, AI experts anticipate a major change from the status quo. Because of this, ChatGPT is not intended for information-seeking on the internet (like a search engine). Instead, the chatbot generates a response using data analysed from substantial portions of training data.
“As opposed to the many links of conventional search engines, ChatGPT can respond to its customers’ questions with a single, concise answer. It also has characteristics that go far beyond those of conventional search engines, including the capacity to create new text, define ideas, engage in back-and-forth communication with the user, among other things “Korinek stated. Even ChatGPT’s developers have discovered emergent features that they were unaware the system had.
On Jan. 23, Microsoft made plans to devote an additional $10 billion in resources to OpenAI, the company behind ChatGPT. The partnership would help both businesses stay on the bleeding edge of “generative AI,” a technology employed in ChatGPT that can learn from vast quantities of data to generate almost any type of content (writing, graphics, music, etc.), all from a simple text command.
Microsoft has a number of consumer-facing tools, like search, that may undergo significant change for users in the years to come. According to The Information, the Seattle-based software behemoth also intends to incorporate ChatGPT’s AI technology into long-established programmes like Word, PowerPoint, and Outlook, in an effort that might revolutionise how over a billion people work and complete daily activities. Integrating it into Outlook, for example, may simply mean instructing the email client to create a message about a given topic.
Microsoft announced the extended agreement in a press statement, saying that it will “use OpenAI’s models across our consumer and corporate products and provide new categories of digital experiences built on OpenAI’s technology.”
Conversational Inquiry
In contrast, Google has been developing such systems for years along with its cutting-edge division DeepMind. However, the search engine giant decided against making them public due in part to worries about unethical activity and how chat platforms can violate social standards. For instance, Microsoft developed the Tay chatbot in 2016 but had to take it down after it started spewing hate speech. With the correct triggers, even ChatGPT, which has guidelines for providing positive and amiable material, may be persuaded to provide distressing replies.
Demis Hassabis, CEO of DeepMind, said in a recent interview with Time that his business is considering developing its own chatbot, named Sparrow, for a “private beta” sometime in 2023. According to a New York Times article citing persons familiar with the situation, Google also intends to display a version of its search engine with chatbot functionality as early as this year.
It’s no secret that, in general, Google search has gotten more conversational over the years. With Google Assistant and knowledge panels in search, the firm has made strides in this area. For years, the corporation has promoted dialogue as the future of search, showcasing its AI systems LaMDA and MUM at its 2021 I/O developer conference.
Microsoft appears to be using OpenAI’s artificial intelligence to outwit Google at its own game. According to The New York Times, Google management declared a “code red” following the launch of ChatGPT. In-house teams have reportedly been reallocated to begin working on AI between now and an anticipated business conference in May, according to the source.
According to Statista, Google’s search engine will continue to dominate the industry in 2022, holding an 84% share of the worldwide search market, compared to Bing’s 9% (although it has increased recently).
A request for comment on this article was not immediately responded to by Google.
How Perceptive Is ChatGPT?
As you’ve undoubtedly already heard, ChatGPT is an advanced chatbot that gained widespread popularity after its initial launch in late November as a no-cost online tool available to everyone with an internet connection. The AI-powered chatbot gained notoriety in part because of its capacity to produce charming poetry, create meal plans, and respond authoritatively to challenging inquiries in a matter of seconds. The technology behind it isn’t technically new, but before ChatGPT, no chatbot has been able to pique the interest of the general public in the same manner. This is primarily attributable to the fact that OpenAI created the ChatGPT phenomena by constructing a chic user experience around the GPT-3.5 language model.
GPT-3.5 is an updated version of GPT-3, which made its debut in 2020 and gained its capabilities through learning from enormous amounts of data and code. GPT-3 contains 175 billion parameters and was trained on 570 terabytes of text, according to Stanford University researchers. (Google’s Dale Markowitz placed the amount of text data at 45 terabytes, “nearly all of the public web,” though.) In comparison, GPT-2, which had 1.5 billion parameters, was almost 100 times smaller.
“The behaviour of the model is substantially altered by the increase in size; the GPT-3 is now capable of carrying out tasks for which it was not specifically taught, such as translating words from English to French, with little to no training data. In GPT-2, this behaviour was mostly missing “In a 2021 post, scientists from Stanford’s Institute for Human-Centered Artificial Intelligence noted.
Korinek stated that the current iteration of ChatGPT “can communicate such information in the consumable form” and likely already knows more about the globe than any one person.
ChatGPT has numerous potential uses, but it also has several restrictions. According to ChatGPT’s detractors, it’s not always obvious where the chatbot gets its data, which might make it difficult for users to accept the conclusions. Additionally, detractors note that ChatGPT will always be vulnerable because of the incompleteness of the data it was trained on, which may include biased or false information.
The shortcomings of the chatbot as it stands have been acknowledged by OpenAI. The software problems with “robustness and veracity,” according to CEO Sam Altman, who said in a December tweet that it would be “a mistake to be depending on it for anything critical right now.”
But don’t expect the AI train to slow down anytime soon.
The key impact of the increased competition, according to Korinek, is that users will have more options and, ideally, better products. “A lot of innovative systems, like ChatGPT, will join the market in 2023,” he continued.
According to reports, the currently under construction GPT-4 has 100 trillion parameters. But according to Altman, a release won’t happen until OpenAI is “sure we can [release] it securely and responsibly.” This was stated in an early January interview with StrictlyVC.
The advanced language model that supports ChatGPT, GPT, was released in four iterations. Altman made an effort to control expectations by stating that “we don’t have AGI” in the fourth version. Artificial general intelligence (AGI) refers to a system that develops its own emergent intelligence independently of the deep learning models that OpenAI presently uses. It’s the type of intelligence that has been portrayed in science fiction works for more than a century and that the critically acclaimed dystopian television programme Westworld helped to promote recently.
In the same interview, Altman stated, “I think [AGI] is sort of what is anticipated of us,” adding that GPT-4 is “going to disappoint” those who hold out that expectation.
