ChatGPT arrived with a jolt, and suddenly, GenAI captivated the public in a manner seen only with the most transformational innovations. This research collection offers explanations of ChatGPT and GenAI, the underlying technology, and Gartner’s view of their impact and potential.
ChatGPT has galvanized public attention since its launch on 30 November 2022. Suddenly, everyone, from CEOs to chefs, marketing executives to musicians, board directors to barbers, was talking about artificial intelligence. Then they began playing with ChatGPT, interacting with a humanlike chatbot that seemed to understand their intentions. It created poetry. It explained complex topics clearly enough for 7-year-olds to understand. It mistakenly wrote believable obituaries for people still alive. It posed an immediate threat to information businesses, and it seemed full of opportunity. It is all of that and more (see Figure 1).
Rarely has Gartner seen such an explosion of public interest and market activity. The market, of course, will settle as players gain market share, combine through acquisitions or disappear altogether. But, unlike previous technology trends, it is easy for everyone in an enterprise to imagine how this technology could be used for basic things, as people can experience it directly. This will make generative AI (GenAI) a durable trend, along the lines of the internet or smartphones.Sure, the hype will burn off as the reality of implementation sets in. But Gartner believes GenAI is here to stay and its impact will continue to grow as people discover new applications for the technology in their daily work and lives. Meanwhile, GenAI will continue to evolve, improving its reliability and finding ways to deal with intellectual property and other challenges.
Some recommended content may not be available as part of your current Gartner subscription.
ChatGPT is an application that sits atop a large language model (LLM) called generative pretrained transformer (GPT). LLMs are trained on billions of words and trillions of parameters. Complex math and enormous compute are required to create these trained models, but they are, in essence, prediction algorithms. In the case of GPT and other LLMs, they predict patterns and sequences of words and generate language that makes sense in response to a question, called a “prompt.”
This generation is characteristic of GenAI and is not limited to text. GenAI can create code, images, video and audio, too. The more reinforcement learning with human feedback (RLHF) is applied during model training, the higher-qualitythe generated responses are (see Figure 2). The humanlike nature of ChatGPT interactions is mesmerizing, and has people camped out in the uncanny valley.
GenAI technologies can generate new, derived versions of content, strategies, designs and methods by learning from large repositories of original source content. GenAI has profound impacts on businesses. They include content discovery, creation, authenticity and regulations; automation of human work; and the customer and employee experience.
The underlying technologies that led to this galvanizing experience include the LLMs themselves (there are hundreds and quickly multiplying), natural language processing and massive computing using GPUs. Of course, the model requires an enormous amount of digitized content, which comes from the public internet, books and other information sources (some of which are, problematically, copyright-protected). Human trends have also converged: our willingness to talk to computers to accomplish tasks, the desire for a better user experience in information retrieval, and increased productivity with less burnout.
Yet, people also fail to distinguish fact from fiction, thanks in part to social media and confirmation bias. This is a problem because these chatbots sometimes make things up. They “hallucinate” and can reflect biases inherent in thetraining data.
▶ Recent advances in GenAI make it the most disruptive set of technologies and capabilities to hit the global market in decades. Boards of directors must become familiar with the competitive opportunities provided by GenAI as well as potential threats to meet enterprises’ evolving oversight needs.
▶ Despite economic headwinds, AI investment is growing, fueled by GenAI advancements. Executive leaders are prioritizing revenue growth and customer experience over cost savings and will push for relatively quick time to production.
▶ Use this research to get answers to your questions on the hot topic of ChatGPT. These questions are distilled from client inquiries, Gartner Peer Connect forums and vendor discussions, and our answers represent Gartner’s take on the topic at a high level.
▶ Due to their scale and wide use-case applicability, foundation models are a major AI advancement, but the major risks they pose aren’t fully understood. Minimize these risks and tap into business value by evaluating the opportunities, benefits and pitfalls these models present.
▶ GenAI can create original media content, synthetic data and models of physical objects to provide breakthrough innovation opportunities. Determine how the multiple use cases for GenAI can impact your organization.
Things have evolved since then to partly address this issue. However, any enterprise serious about using GenAI still needs the familiar mix of information protection and governance techniques, such as encryption, identity and access management, and information classification. Other risks of GenAI include:
Transparency risks: LLMs are not explainable. Even the vendors don’t understand everything about how they work internally.
Reliability risks: The output of LLMs contains untruths as well as truths, and these may be retrieved or synthesized from the training data or the context of prompts. Outputs generated by ChatGPT and GenAI should be validated and moderated to ensure quality assurance is integral to use.
Bias risks: LLM training is based on potentially biased data selectively sourced from the web and books. Enterprises must have in place policies or controls to detect biased outputs and deal with them in a manner consistent with company policy and any relevant legal requirements.
IP and copyright risks:In the specific case of the ChatGPT application (not the direct API access model), there are currently no verifiable data governance and protection assurances regarding confidential enterprise information. Users should assume that any data or queries they enter into ChatGPT will become public information. Enterprises should take steps to avoid inadvertently exposing enterprise IP. Users must also filter out copyrighted materials in ChatGPT outputs to which OpenAI did not have the rights, creating potential legal risks for the user. The lack of reliable source references makes this a serious challenge.
Cyber and fraud risks: Enterprises must prepare for the use of GenAI systems for cyberattacks and fraud, such as deepfakes of enterprise products and communications. Enterprises should confer with their cyberinsurance providers to verify the degree to which AI-related breaches are covered by their existing policies.
Sustainability risks: GenAI uses significant amounts of electricity and water. Enterprises that invest in GenAI should encourage the executive team to choose vendors that reduce power consumption and leverage high-quality renewable energy to mitigate the impact on sustainability goals.
Many enterprises are quickly coming to terms with these risks, both to their IP and to their business models. It is essential to create and communicate a clear usage and governance policy that determines appropriate use of GenAI tools, protection of data and acceptable use cases.
▶ General counsel, along with leaders in privacy, IP and information security, can share examples of risks from our “interview” with the tool while creating policies to help the enterprise keep up.
▶ No other technology has had a faster uptake than ChatGPT, and it will be followed by other LLM applications, multiplying its profound impact on the business landscape. Establish best practices for navigating the new and emerging ethical implications of this groundbreaking technology.
▶ LLMs like ChatGPT can save time and automate tasks, but they also pose risks via multiple vectors: use by outside actors, ungoverned employees and enterprises. Chief audit executives must assess this technology’s capabilities as well as risks to their organizations’ AI governance.
▶ ChatGPT raises many trust, risk and security threats to enterprise users. To avoid exposing enterprise confidential information, manually review all outputs and set clear organizational policies for use until enterprise-grade security and risk controls are implemented.
▶ To craft an effective policy, the general counsel must consider risk tolerance, use cases and restrictions, decision rights, and disclosure obligations. Get insights on how to set up guardrails, including sample language from the city of Boston’s GenAI policy.
▶ Many organizations have not issued guidance on employee use of GenAI like ChatGPT, but most are working on it. HR leaders can use the examples explored in this research to review and inform their position toward GenAI.
Exploratory investigations done by end-user organizations and vendors reveal use cases manifest in the user experience, as well as deep within the workings of applications and their programmatic interfaces. GenAI will play a foundational role in many processes and applications by composing existing and newly emerging models and techniques, with myriad use cases arising.
Augmented authoring: It generates or drafts copy, reformats text, changes style and tone (such as professional, informal and simplified tone), simplifies, and so on.
Augmented reading: It summarizes, extracts facts, answers questions and so on.
Translation: It translates languages, programming languages, data formats and so on.
Knowledge management: It retrieves and synthesizes information, classifies content and data assets, and so on.
Programming: It generates code or queries and synthetic data for training, testing or simulations.
Potential use cases must be evaluated for value and feasibility to determine priority. Gartner is helping clients assess and prioritize use cases in conjunction with our consultants and executive partners. We are building out research to examine use cases as they relate to industries, business functions, roles and markets — most notably through our Use-Case Prisms.
▶ ChatGPT is a GenAI foundation model that has a great number of possible uses to the enterprise. Our downloadable Tool lists and profiles use cases for ChatGPT to help end users understand what is possible with the model and the restrictions inherent within its use cases. It also provides examples of potential enterprise uses.
▶ Innovation teams incorporating GenAI in their work get strong results fast. To harness this for an accelerating arms race, CIOs should combine human and artificial intelligence to shorten and improve innovation and chart a path that could lead to near-autonomous innovation.
▶ ChatGPT has created significant interest in the potential of LLMs to improve healthcare efficiency, experience and outcomes. CIOs must separate the hype from reality by learning the potential use cases, limitations and risks associated with deployment of this technology.
▶ GPT-4 has exploded onto the scene in healthcare and life sciences, igniting CIOs’ innovative new thinking, excitement and concern about a future of healthcare fundamentally reshaped by GenAI. This research will help CIOs understand how to assess opportunities and risks for their organizations.
▶ Although there is hype around ChatGPT, it is only one solution among the larger scope of GenAI opportunities. This research shows where the insurance industry can leverage GenAI’s unique capabilities, multiple use cases and great long-term value appropriately in both life and P&C insurance.
▶ Many large companies are planning to use ChatGPT to support customer service and analytics visions. Insurance CIOs must know the use cases, benefits and risks around it to guide their decisions about ChatGPT.
▶ Bank CIOs can learn the implications of foundation models for their digital strategies and see what tech investments their peers are prioritizing to keep up with this evolving technology.
▶ In April 2023, only 7% of banking executives reported no plans to incorporate ChatGPT or other GenAI into the business, down steeply from 46% in February. Learn about why banking executives can no longer “wait and see” when it comes to GenAI, and how they plan to incorporate it.
▶ Thriving in times of economic and social volatility requires organizations to find new ways to scale productivity, invest in innovation, and grow and retain revenue. Investment management CIOs must adopt, trial or assess these top technology trends to influence priorities, strategies and execution.
▶ Brand leaders eager to leverage ChatGPT in campaigns, messaging and other public executions must use caution. Consumer awareness of ChatGPT is emerging, and the perception of brand use for external purposes creates brand trust concerns and reputational risk, regardless of actual use cases.
▶ As hype around ChatGPT-like GenAI solutions continues, HR leaders have various considerations around their goals, use cases, investment plans, concerns and guidance. HR leaders can use this report to benchmark and subsequently decide what role GenAI may play in their HR strategy.
▶ LLMs offer significant opportunities and several potential risks for enterprise users. Legal and compliance leaders should assess the level of risk exposure and build appropriate measures to steer a responsible use of LLMs — and other GenAI tools — within the enterprise.
▶ GenAI will have a significant impact on jobs and workers, but impact will vary substantially based on the labor market and other demand drivers. Executive leaders, along with their HR partners, should use this research to structure and adapt their strategies to future-proof their workforce.
The pilots come in three categories, with the second being the most common:
▶ When organizations begin experimenting with GenAI, they often neglect its transformative nature and the operationalization challenges it entails. IT leaders responsible for AI can follow the five steps in this research to pilot GenAI and start delivering business value.
▶ To make informed decisions and derive value, CTOs need to understand the various approaches. Here, we compare the deployment approaches and provide a decision framework for choosing one over the other.
▶ Data and analytics leaders struggle in determining how to implement LLMs. These downloadable slides provide architecture guidance on a diverse set of design patterns, ranging from out-of-the-box use of ChatGPT and LLMs to complex and customized solutions.
▶ Data and analytics leaders must assess the potential benefits and cost of new GenAI investments. Experimentation can be done inexpensively for most use cases, but this research provides a decision framework for assessing and realizing value from enterprise GenAI initiatives.
▶ OpenAI ChatGPT is a tuned and closed service based on the GPT model. IIt is leveraged by Microsoft Azure OpenAI Service. Use this report to understand how to best leverage this new technology.
▶ This research provides an overview of AI and Gartner’s AI-related research and analyst resources. Use this top-level document to identify areas for deeper investigation and develop AI plans and strategies to create an effective AI ecosystem.
▶ Data and analytics technical professionals must define their roles and work together as part of an AI team. This research defines core AI and machine learning (ML) roles, skills and responsibilities, helping to align the right skills to the required roles in advanced analytics initiatives.
▶ Microsoft is rapidly rolling out Copilot, everyday AI services based on OpenAI’s GPT across its application and cloud portfolios, including Microsoft 365. Most organizations are unprepared for the changes Copilot will unleash.
▶ Economic conditions are reducing Google’s advertising results and slowing its ability to deepen its workforce commitments. Google Cloud leads in growth, while analytics and AI strive to become a primary option. Expect Google to expand its presence in the enterprise.
▶ GenAI is a game-changing technology. Tech CEOs must have a compelling offering strategy that looks beyond adoption of GenAI while also leveraging it for competitive advantage and being ready for a significant pivot in their business.
▶ Sales engagement applications help sales operations leaders streamline daily digital seller workflow tasks, guide seller decision making with AI/ML and provide a seller-centric user experience for SFA/CRM. Use this Market Guide to evaluate best-fit vendor options to optimize B2B sales productivity.
Gartner has historical foresight on this topic, such as the 2020 Gartner Data & Analytics Summit with the theme of Rewire Your Culture for an AI-Augmented Future. See also and .