|
Have you ever found yourself searching through many pages of results on a search engine to find the solution to a hard question? Say you want to find out whether a vegetarian diet is acceptable for your dog. Your research trip may begin with getting onto Google and putting "is a veg diet healthy for dogs" into the search box and then trying to make sense of the legion of produced links. By the time you uncover an answer, you've buried considerably more time than you'd anticipated into reading through publications, reports and their sources.
In the not-so-distant future, finding the solution to a hard issue may not be such a tiresome, mind-numbing procedure. Microsoft is apparently merging the AI software that drives ChatGPT into its Bing search engine in a move that might revolutionize search as we know it. More precisely, Bing could have the capacity to deliver up a search experience that's better than Google, according to AI experts, and possibly challenge the search giant's decades-long supremacy.
"ChatGPT is the first new technology in more than a decade that may actually revolutionize search and that may, at least in theory, upend Google's market supremacy," said Anton Korinek, an AI researcher and professor of economics, at the University of Virginia. "What the technology accomplishes is that it enables users to communicate with their computer in a far more natural and conversational fashion than conventional search."
At this moment, we don't know what Bing's AI-driven search results would look like precisely. Microsoft refused to comment for this report. However, AI experts foresee a major break from the status quo in terms of how a search engine provides an answer and how consumers interact with it. After all, ChatGPT is not meant to scan the internet for information (like a search engine) (like a search engine). Instead, the chatbot leverages information studied from enormous swathes of training data to construct a response.
"ChatGPT can provide its users with a single unambiguous response compared to the myriads of links of standard search engines. It also has features that go much beyond standard search engines, such [the capacity] to produce new content, explain ideas, have a back-and-forth discussion between the user and the system, and so on," added Korinek. "People still uncover emergent capabilities that even the developers of ChatGPT were not aware that the system possessed."
|
| Microsoft CEO Satya Nadella and OpenAI CEO Sam Altman at the Microsoft campus in Redmond, Washington, in July 2019. |
Microsoft revealed intentions on Jan. 23 to pour further resources into OpenAI, the originator of ChatGPT, to the tune of $10 billion. The merger would help maintain both firms at the bleeding edge of what's known as generative AI, a tech utilized in ChatGPT that can learn from huge amounts of data to produce almost any content type (writing, photos, audio and so on) just from a text input.
Search is only one of a spectrum of consumer-facing products in Microsoft's portfolio that might possibly alter dramatically for users in the coming years. According to a report by The Information, the Seattle-based software giant also has ambitions to incorporate ChatGPT's AI tech into long-established products like Word, PowerPoint and Outlook in an initiative that might transform how more than a billion people work and do everyday activities. For instance, incorporating it into Outlook may mean just requesting the email program to create a message on a given subject.
"Microsoft will deploy OpenAI's models across our consumer and business products and launch new categories of digital experiences based on OpenAI's technology," the firm stated in a news statement announcing the extended relationship.
Conversational search
For its part, Google and its cutting-edge subsidiary DeepMind have been working on comparable algorithms for years. The search giant elected not to reveal them to the public, however, in part amid worries about unethical behavior and how chat platforms frequently undermine social standards. For instance, in 2016 Microsoft produced a chatbot named Tay that it was compelled to take down when it spewed forth hate speech. Even ChatGPT, which includes guidelines to promote good and pleasant material, may be fooled into creating distressing answers utilizing the correct cues.
However, DeepMind CEO Demis Hassabis revealed in an interview with Time this month that his business is exploring its own chatbot, named Sparrow, for a "private beta" sometime in 2023. Google also aims to display a version of its search engine with chatbot characteristics as early as this year, according to a New York Times story citing persons familiar with the subject.
It's no secret that Google search has gotten more conversational in general over the years. The business has made headway in this area with the Google assistant and with knowledge panels in search, and for years has touted dialogue as the future of search, demoing its AI systems LaMDA and MUM at its 2021 I/O developer conference.
Leveraging OpenAI's artificial intelligence appears to be how Microsoft is seeking to beat out Google at its own game. In the aftermath of ChatGPT's leak, Google management declared a "code red," according to The New York Times. The article stated internal teams have been allocated to commence work on AI between now and a planned business meeting in May.
Still, Google's search engine now remains the uncontested industry leader as it has for decades, holding 84% of worldwide search market share, compared to Bing's 9% (although it has increased in recent years) in 2022, according to Statista.
Google didn't immediately reply to a request for comment on this story.
|
| Chatbots got a lot more complex with OpenAI's ChatGPT tool. |
How clever is ChatGPT?
As you've surely heard by now, ChatGPT is a smart chatbot that went viral internationally following its initial release in late November as a free online tool available to anybody with an internet connection. The AI-powered chatbot grabbed headlines owing in part to its capacity to churn out charming poetry, prepare meal plans and deliver authoritative answers to hard inquiries within seconds after being instructed. The tech underpinning it isn't exactly fresh new, but no chatbot had yet managed to catch widespread curiosity in the manner that ChatGPT did. That's primarily because OpenAI designed a nice user experience around the GPT-3.5 language model, and that's the phenomena we know as ChatGPT.
GPT-3.5 is an enhanced version of GPT-3, which debuted in 2020 and which learnt from enormous tracts of data and code to help it reach its skills. According to researchers at Stanford University, GPT-3 was trained on 570 terabytes of text and includes 175 billion parameters. (Google's Dale Markowitz, however, estimated it at 45 terabytes of text data, "containing practically all of the public web.") For contrast, its predecessor, GPT-2, was nearly 100 times smaller, with 1.5 billion parameters.
"This increase in size substantially modifies the behavior of the model – GPT-3 is able to execute tasks it was not expressly trained on, such translating words from English to French, with few to no training data. This tendency was mainly missing in GPT-2," researchers from Stanford's Institute for Human-Centered Artificial Intelligence stated in a 2021 post.
"The current version of ChatGPT probably already knows more about the world than any given person, and it can communicate that information in palatable form," added Korinek.
For all the potential ChatGPT has, there are almost as many constraints. Critics of ChatGPT claim it's not always apparent where the chatbot is taking information from, which might make it difficult for consumers to trust the findings. Skeptics further point out that ChatGPT will always be undercut by the poor quality of the data it was trained on, including biased information or disinformation.
OpenAI has recognized the chatbot's shortcomings in its present version. CEO Sam Altman remarked in a December post on Twitter that the product problems with "robustness and veracity" and that it would be "a mistake to be depending on it for anything significant right now."
But don't expect for the AI bandwagon to slow down.
"There will be a lot of new systems like ChatGPT that will reach the market in 2023, and the major meaning of the subsequent rivalry is that customers will have more choice and, ideally, better goods for consumers," continued Korinek.
GPT-4, which is in development, is estimated to have 100 trillion parameters. But a release is not likely to take place until OpenAI is "sure we can [release] it securely and responsibly," Altman said in an interview with StrictlyVC in early January.
Altman has sought to downplay expectations of that fourth generation of GPT, the complex language model that supports ChatGPT, claiming "we don't have AGI." AGI stands for artificial general intelligence, meaning a technology having its own emergent intellect as opposed to depending on the deep learning models now utilized by OpenAI. It's the type of intelligence that has been portrayed in science fiction novels for more than a century and was popularized in recent years by the award-winning dystopian program Westworld.
"I believe [AGI] is kind of what is anticipated of us," Altman said in the same interview, adding that GPT-4 is "going to disappoint" anyone who hold out that expectation.
Editors' note: CNET is utilizing an AI engine to construct certain personal finance explainers that are reviewed and fact-checked by our editors. For more, check this post.