Chatbot platforms like ChatGPT and GPT-3 are valuable tools for automating functions, creating creative ideas, and even writing new code and providing fixes for broken applications, but in Before businesses adopt it, there are some precautions that need to be taken.
For businesses, chatbots like ChatGPT have the potential to automate routine tasks or enhance complex communications, such as creating email sales campaigns, modifying computer code, or improving customer support.
Research institute Gartner predicts that by 2025, the global market size of artificial intelligence software will reach US$134.8 billion, and the market growth rate is expected to increase from 14.4% in 2021 to 31.1% in 2025, far from outpacing the overall growth of the software market.
A large part of this market will be chatbot technology, which uses artificial intelligence and natural language processing to respond to user queries, provide human-like answers, and more sophisticated programs can answer follow-up questions and can Modify for specific business purposes.
In a survey report released recently, IDC elaborated on the possible uses of ChatGPT and its basic language model GPT-3 (also including GPT3.5 and GPT4). These models can be customized.
ChatGPT, launched by OpenAI in November last year, quickly became popular around the world, with 1 million users in the first five days of launch, because of its sophisticated way of generating in-depth, human-like responses.
Currently, ChatGPT is most commonly used as an “out-of-the-box” text-based web chat interface. GPT-3 currently does not provide API access functionality. Microsoft plans to provide an API for its version of Azure OpenAI ChatGPT, which will be available soon.
Businesses can use out-of-the-box methods to augment or create content, manipulate text in emails to soften language or adopt a specific tone, and summarize or simplify content. "This can be achieved with limited investment," Gartner explains in the report.
ChatGPT and GPT- 3 (which stands for Generative Pre-trained Transformer) are all machine learning language models trained by OpenAI, an artificial intelligence research laboratory based in San Francisco. While both ChatGPT and GPT-3 can generate human-like textual responses to queries, they are not equal in level of sophistication.
Muhammad, senior solutions architect at TripStax, said one of the main differences between ChatGPT and GPT-3 is their size and capacity.
He wrote in a blog post: "ChatGPT was specifically designed for chatbot applications, GPT-3 is more general and can be used for a wider range of tasks. This means that ChatGPT may be more effective To generate responses in a conversational context, GPT-3 may be better suited for tasks such as language translation or content creation."
ChatGPT cannot be customized at this time because there is no access to the language model it is based on. ChatGPT is not an open source software application. However, OpenAI already provides the GPT-3 model, as well as other large language models (LLM). Large language models (LLMs) are machine learning applications that can perform many natural language processing tasks.
Gartner said: “Because the underlying data is target-specific, having more control over the entire process may yield better results. Although this approach requires significant skills, data management and funding, but the emergence of a market for third-party, purpose-built models may make this option increasingly attractive."
For example, Microsoft Corp.'s OpenAI service leverages ChatGPT to provide business and application developers with Provides a way to take advantage of new technology. Microsoft's latest improved Bing search engine uses GPT-4 (the latest version of OpenAI).
ChatGPT is based on a smaller text model that can accommodate about 117 million parameters. Muhammad noted that GPT-3 was trained on 45TB of text data, and its capacity is much larger, with 175 billion parameters.
Since ChatGPT has not been connected to the internet before, it will occasionally produce incorrect answers. According to the company's FAQ, OpenAI has limited visibility into global events beyond 2021 and may occasionally produce harmful instructions or biased content.
Bern Elliot, vice president and distinguished analyst at Gartner, said: "ChatGPT is an application that consists of pre-preparation and post-screening, as well as a customized version of GPT-3.5. People submit questions and any additional information, saying for hints. Although custom GPT-3 is not accessible in ChatGPT, the way the question is asked has an important impact on the quality of the results."
This is often referred to as "hint engineering" and can be used in any large language model Completed on. In many cases, users also have access to underlying large language models (LLMs) such as GPT-3.
Gartner said that fundamentally, ChatGPT can be used to improve content creation and conversion automation while providing a fast and engaging user experience.
The easiest way to use ChatGPT is the question and answer method. For example, "How many kilometers does it take to drive from Boston to San Francisco?"
ChatGPT can also be used to create written content, or enhance already written content by softening or specializing the language to give it a different tone.
Gartner stated in the report: "ChatGPT can generate 'draft' text of the required length and style in a variety of ways for users to review. Its specific uses include draft marketing descriptions, letters of recommendation, papers, Manuals or instructions, training guides, social media or news posts."
Gartner's Elliot added: "Creating email sales campaign materials or providing answers to account agents is a reasonable approach."
Chatbot technology can also provide summaries of conversations, articles, emails, and web pages.
Another use for ChatGPT and GPT-3 is to improve existing customer service chatbots so that they provide more detailed and human-like responses.
These platforms can also improve customer intent recognition, summarize conversations, and answer customer questions. Doing so requires enterprise scenarios, service descriptions, permissions, business logic, forms of tone, and even brand tone, which all need to be added to the GPT-3 language model.
Sales and marketing can also use ChatGPT and GPT-3 to provide recommendations and product descriptions to potential customers on the website or via chatbots. Likewise, chatbot platforms need to be customized to the enterprise environment.
Chatbots are also used as personal assistants for managing schedules, summarizing emails, composing emails and responses, and drafting common documents.
In education, chatbots can be used to create personal learning experiences, much like a tutor. And, in healthcare, chatbots and apps can provide simple language descriptions of medical information and treatment recommendations.
Elliot explained that the GPT-3 model inside the ChatGPT service cannot be modified individually, but users can obtain the basic GPT-3 model and modify it individually to use it in the chatbot. Used in the engine (ChatGPT application not required). GPT-3 models can be used simply like other large language models (LLMs).
For example, users can add data and tuning parameters for a GTP-3 model or dataset. The way people submit questions to these models can also be affected by the wording used when asking. "So, just-in-time engineering is also useful," Elliot said.
Gartner said in the report: "Although this approach requires significant skills, data management and capital, the market for third-party purpose-built models is "The emergence of ChatGPT and GPT-3 may make this option increasingly attractive."
While the use of ChatGPT and GPT-3 for software code generation, translation, interpretation, and verification is just emerging, it promises to enhance the development process. Gartner says it will most likely be used within an integrated development environment (IDE).
For example, a developer can type into the search box: "This code doesn't work the way I expect, how do I fix it?" OpenAI noted that while the first response given by ChatGPT was not unlikely to solve the problem, but through follow-up questions on the responses, a solution can be devised.
ChatGPT can also write code, convert code from one programming language to another, correct error codes, and interpret codes.
Elliot says, "It's valid to suggest alternatives to software code or identify coding errors, but don't let ChatGPT 'fix' the code, instead suggest what needs to be checked."
Gartner warns that relying on ChatGPT is risky because many users may not understand the data, security and analytics limitations.
One of the biggest concerns for businesses, the company said, is that ChatGPT may employ natural language to generate content with little value, or worse, untrue statements. Users are mandated to check the output for accuracy, appropriateness and practical usability before accepting any results.
Using chatbots can also expose confidential and personally identifiable information (PII), so it is important for businesses to be careful about the data used for chatbots and avoid including confidential information. It’s also critical that businesses work with vendors that offer strong data usage and ownership policies.
OpenAI CEO Sam Altman warned users in a tweet in December that ChatGPT’s functionality was very limited and that relying on it to do anything important now would be a mistake.
Altman wrote, "This is a preview of progress, and we have a lot of work to do in terms of robustness and authenticity." He added that ChatGPT is best suited for providing "creative inspiration."
Gartner generally agrees with this view in its report, noting in the report, "Recognize that this is a very early stage and is a hyped technology with potentially significant uses, so Continuous improvement all the time."
The company recommends that enterprises encourage "out-of-the-box" thinking about workflows, define guidelines around the use and governance of AI, and establish a working group to provide guidance to chief information officers and chief executives. Executive Report.
Elliot recommends that users adopt Microsoft's Azure Open Service ChatGPT instead of OpenAI's enterprise-grade ChatGPT, he said, "because Microsoft provides enterprise security and compliance controls related to other Microsoft products. If you plan to use confidential information, Then use Azure."
Microsoft said it plans to enable security and privacy policies on Azure OpenAI like it does for other Azure services.
Finally, Elliot said, don’t let employees ask OpenAI ChatGPT questions that might reveal corporate secrets. Businesses need to issue clear policies to educate employees about the inherent risks associated with ChatGPT.
OpenAI was founded in 2015 and is supported by investors such as Elon Musk, AWS, Infosys, and YC Research. Altman became CEO of OpenAI in 2019.
Other early investors include Microsoft, which invested $1 billion in OpenAI in 2019 and recently announced plans to invest billions more. Microsoft also announced that its Bing search engine is being upgraded to use GPT-4, the latest version of the artificial intelligence language model built by OpenAI.
The announcement sparked a search chatbot war between Microsoft and Google, with Microsoft hoping the use of GPT-4 would allow Bing to leapfrog Google's long-dominant search engine. Google has also released its own chatbot technology, Bard, a conversational artificial intelligence service powered by a technology called Language Model for Conversational Applications (LaMDA).
Preply, a global language learning platform, released the results of a study that compared the intelligence of Google search engine and ChatGPT. Preply organized a so-called "communications expert panel" to evaluate 40 intelligence challenges for each AI platform.
The results of the evaluation show that ChatGPT defeated the Google search engine 23 to 16, and there was also a draw. However, Google performs better on basic questions and queries where information changes over time.
The above is the detailed content of How Enterprises Use ChatGPT and GPT-3. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
