Generative AI holds huge economic potential for the global economy

But it is their ability to write text, compose music, and create digital art that has garnered headlines and persuaded consumers and households to experiment on their own. As a result, a broader set of stakeholders are grappling with generative AI’s impact on business and society but without much context to help them make sense of it. However, a collaborative approach from government, industry and education providers is essential. “Skilling programs exist today in pockets across Asia, but too many people are severely underserved because of race, gender, geography, displacement, or other barriers,” Mazhari says. The fact that 61% of students do not receive any digital literacy education at school in ASEAN countries means it is imperative that swift action is taken now—to ensure this technology’s economic impact in the future.

• As a result, companies don’t have to stress about extra costs resulting from job-related accidents, and employees can focus on other lower-risk tasks.
• While the generative AI gold rush has sparked a surge in spending on both buying large language models (LLMs) and building their own, companies are beginning to realize that the technology is failing to live up to the promises they were sold on.
• Second, previous papers outlining research questions about the effect of AI on marketing have focused on either the consequences on firms’ activity (Davenport et al., 2020; Huang & Rust, 2021, 2023) or consumer response (Puntoni et al., 2021).
• They attributed this to the tools’ ability to automate grunt work that kept them from more satisfying tasks and to put information at their fingertips faster than a search for solutions across different online platforms.

This is an essential part of what’s

called a “neural network architecture.” The discovery of new architectures has been an important area of AI

innovation since the 1980s, often driven by the goal of supporting a new medium. But then, once a new

architecture has been invented, further progress is often made by employing it in unexpected ways. Empirically, we know how they work in

detail because humans designed their various neural network implementations to do exactly what they do,

iterating those designs over decades to make them better and better. AI developers know exactly how the

neurons are connected; they engineered each model’s training process. Yet, in practice, no one knows exactly

how generative AI models do what they do—that’s the embarrassing truth.

From this perspective, GenAI necessitates complementary market-based assets to realize its full potential. Having reviewed how GenAI models are trained and how they produce new content, we next describe the capabilities and current limitations of these models. Huang and Rust (2023) conceptualize GenAI as feeling AI, namely a type of AI that can communicate and interact with humans based on an emotional understanding derived from analyzing emotion data. They argue that these models differ from previous AI models, which were better suited for mechanical AI (i.e., performing repetitive tasks) and thinking AI (i.e., supporting analytical decisions) (Huang & Rust, 2021). We embrace their view of GenAI as a novel type of AI; to complement this view, we focus on GenAI’s capability to create new content.

Encoders compress a dataset into a dense representation, arranging similar data points closer together in an abstract space. Decoders sample from this space to create something new while preserving the dataset’s most important features. The high-tech/telecom industry accounted for 38% of the risk assessment sector, underscoring the need for risk control in the industry.

Generative AI refers to conversational AI tools like OpenAI’s ChatGPT released in November, which impressed the world with its wide-ranging abilities including creating content, generating music, and writing code. Since the release of ChatGPT in November 2022, it’s been all over the headlines, and businesses are racing to capture its value. Within the technology’s first few months, McKinsey research found that generative AI (gen AI) features stand to add up to $4.4 trillion to the global economy—annually. Put simply, business leaders need to emerge themselves now in generative AI (anyone who can ask questions can use the technology) and be prepared to learn continuously. We found significant improvements in worker productivity as measured by the number of customer issues workers were able to resolve per hour. Within four months, treated agents were outperforming nontreated agents who had been on the job for over twice as long.

Despite the immense creativity generative AI provides, ethical considerations regarding accountability, bias, and privacy arise. For instance, can companies fully trust an AI tool to process their employees’ sensitive data safely? These matters should be addressed early on and companies must devise plans to effectively treat the gray areas. The PwC study “Digitalisation in Finance and Accounting 2023” shows that almost 60 per cent of the companies surveyed are currently driving forward a comprehensive transformation of their finance function. Staff shortages and regulatory requirements are putting increasing pressure on the integration of technologies.

Historically, technology has been most effective at automating routine or repetitive tasks for which

decisions were already known or could be determined with a high level of confidence based on specific,

well-understood rules. Think manufacturing, with its precise assembly line repetition, or accounting, with

its regulated principles set by industry associations. To suggest an admittedly extreme example, generative AI might assist an

organization’s strategy formation by responding to prompts requesting alternative ideas and scenarios from

the managers of a business in the midst of an industry disruption. Another difference worth noting is that the training of foundational models for generative AI is “obscenely

expensive,” to quote one AI researcher. Say, $100 million just for the hardware needed to get started as

well as the equivalent cloud services costs, since that’s where most AI development is done. The latest generative AI applications can perform a range of routine tasks, such as the reorganization and classification of data.

By automating time-consuming tasks, creating high-value content, and targeting specific buyer personas, companies save money while creating new job roles. Generative artificial intelligence (AI) systems are trained on large data corpora to generate new pieces of text, images, videos, and other media. There is growing concern that such systems may infringe on the copyright interests of training data contributors.

But a much smaller share of respondents report hiring AI-related-software engineers—the most-hired role last year—than in the previous survey (28 percent in the latest survey, down from 39 percent). Roles in prompt engineering have recently emerged, as the need for that skill set rises alongside gen AI adoption, with 7 percent of respondents whose organizations have adopted AI reporting those hires in the past year. Recent efforts in machine learning have primarily focused on minimizing the likelihood of creating copyright-infringing content by generative AI models.

History of Generative AI

For example, Citibank has partnered with data science leader Feedzai to use AI technology to detect and prevent fraudulent activity and protect trillions of dollars in daily transactions, greatly improving the security of customer funds. The online survey was in the field April 11 to 21, 2023, and garnered responses from 1,684 participants representing the full range of regions, industries, company sizes, functional specialties, and tenures. Of those respondents, 913 said their organizations had adopted AI in at least one function and were asked questions about their organizations’ AI use. To adjust for differences in response rates, the data are weighted by the contribution of each respondent’s nation to global GDP. AI high performers are expected to conduct much higher levels of reskilling than other companies are.

The report also warned that the benefits of AI could be unevenly distributed, however, with some workers and regions experiencing more significant job displacement than others. Gen AI creates personalized recommendations for movies, TV shows, and music based on individual preferences in the entertainment industry. This technology can foster the same efficiency and accuracy that it does in other industries, making it a potential cost-saver for media companies.

Generative AI and Its Economic Impact: What You Need to Know

The study looked at 16 business functions and examined 63 use cases where generative AI can help ease business challenges. The technology, among other things, supports customer interactions, creates creative content for marketing and sales, and generates computer code based on natural language. Generative AI can substantially increase labor productivity across the economy, but that will require investments to support workers as they shift work activities or change jobs. Generative AI could enable labor productivity growth of 0.1 to 0.6 percent annually through 2040, depending on the rate of technology adoption and redeployment of worker time into other activities.

For instance, face recognition programs trained with images of people from a particular race will probably stumble upon errors when trying to identify other races. Additionally, based on the language and vocabulary humans use to teach generative AI, the latter may form gender, ethnicity, and race bias. We’ve already established the economic potential of generative AI and the risks it possesses. However, is there an even bigger Artificial Intelligence challenges you should know about?

Today’s generative AI models produce content that often is indistinguishable from that created by humans. Extremists on opposite sides

of the debate have said that the technology may ultimately lead to human extinction, on one side, or save

the world, on the other. Enterprises

should be concerned with the ways in which generative AI will drive changes in work processes and job roles,

as well as the potential for it to inadvertently expose private or sensitive information or infringe on

copyrights. It’s important to note that generative AI is not a fundamentally different technology from traditional AI;

they exist at different points on a spectrum. Traditional AI systems usually perform a specific task, such

as detecting credit card fraud. This is partly

because generative AI tools are trained on larger and more diverse data sets than traditional AI.

Percentage of financial services using AI on a daily basis, by business area, 2022

Traditionally, they would need to consolidate that data as a first

step, which requires a fair bit of custom software engineering to give common structure to disparate data

sources, such as social media, news, and customer feedback. “Over the next few years, lots of companies are going to train their own specialized large language models,”

Larry Ellison, chairman and chief technology officer of Oracle, said during the company’s June 2023 earnings

call. Generative AI took the world by storm in the months after ChatGPT, a chatbot based on OpenAI’s GPT-3.5 neural

network model, was released on November 30, 2022. GPT stands for generative pretrained transformer, words

that mainly describe the model’s underlying neural network architecture.

This process allows models to be grounded in vectorized external data to improve outputs, dramatically reducing the need for additional rounds of training and fine-tuning on off-the-shelf models. “Training is the most resource-intensive and costly process, requiring significant compute and power. But, once a mode is trained, the cost of fine-tuning and inference should be lower,” says Łukasz Mądrzak-Wecke, head of AI at enterprise digital product consultancy Tangent, in conversation with ITPro. Nvidia boss Jensen Huang caused a stir earlier this year when he hinted at the death of coding.

We’ve all become familiar with the term ‘hallucinations’, used to describe the confident falsehoods LLMs regularly produce, and efforts to reduce these have only gone so far to date. Above all the myths spread since 2022 about generative AI, those that imply LLMs can outpace human output or are a step towards artificial general intelligence (AGI) straight out of science fiction, are the most damaging. Sign up today to receive our FREE report on AI cyber crime & security – newly updated for 2024.

A good instruction prompt will deliver the desired results in one or two tries, but this often comes down to placing colons and carriage returns in the right place. This ability to generate novel data ignited a rapid-fire succession of new technologies, from generative adversarial networks (GANs) to diffusion models, capable of producing ever more realistic — but fake — images. In Figure 3, we explored the SRS distribution for prompts requesting content generation from multiple data sources. Notably, for the WikiArt dataset, prompts asked the generative model to blend styles from multiple artists. The SRS effectively recognized and rewarded the contributions of data sources integrated into the generated artworks, showcasing the framework’s capability to discern and value diverse data source inputs to generate content. Our framework not only tackles copyright disputes but also addresses scenarios where multiple entities, each holding a private dataset, seek to jointly train a generative AI model with the objective of generating revenue from its application.

This is particularly true when human expertise and ingenuity is paired with deep understanding of how to use these programs and effectively harness their capabilities. Our previous discussion suggests that GenAI has the potential to fundamentally transform the value of key marketing capabilities, spanning from new product development to market information management to communication (Vorhies & Morgan, 2005). Hence, GenAI could redefine the essence of marketing’s contribution to a firm’s sustainable competitive advantage.

Banking, a knowledge and technology-enabled industry, has already benefited significantly from previously existing applications of artificial intelligence in areas such as marketing and customer operations.1“Building the AI bank of the future,” McKinsey, May 2021. Some of this impact will overlap with cost reductions in the use case analysis described above, which we assume are the result of improved labor productivity. To grasp what lies ahead requires an understanding of the breakthroughs that have enabled the rise of generative AI, which were decades in the making. For the purposes of this report, we define generative AI as applications typically built using foundation models.

The productivity potential of GenAI

Therefore, the economic potential of generative ai becomes visible as it helps businesses retain their workforce and improve people’s experiences. Many organizations across the globe are now using AI tools to create content for recruitment as HR benefits highly during acquisition and onboarding. They ask the algorithm to create job postings based on skills, keywords, and older listings. In advanced cases, companies may design avatars for each candidate and provide personalized feedback. A study by the World Economic Forum found that adopting AI could lead to a net increase in jobs in some industries, particularly those that require higher levels of education and skills.

GenAI’s impact can be likened to a black hole’s event horizon – unpredictable and potentially extreme. My research suggests successful companies navigate this uncertainty through strategic investments in AI projects, balancing performance gains with continuous learning. Tesla’s approach to autonomous driving, for example, leverages incremental advancements to build expertise. These are just a few examples of how intelligent technologies are driving meaningful and positive changes in the biopharma industry. Employed correctly, they can create a significant competitive advantage at each step of the value chain.

As an example of how this might play out in a specific occupation, consider postsecondary English language and literature teachers, whose detailed work activities include preparing tests and evaluating student work. With generative AI’s enhanced natural-language capabilities, more of these activities could be done Chat GPT by machines, perhaps initially to create a first draft that is edited by teachers but perhaps eventually with far less human editing required. This could free up time for these teachers to spend more time on other work activities, such as guiding class discussions or tutoring students who need extra assistance.

They point out that firms’ repeated use of GenAI will fundamentally alter marketing capabilities, for instance by making some of them obsolete or less valuable. Also, they expect that consumers’ skills and preferences will likely change as an effect of widespread GenAI use. Thus, they fear that firms will necessarily face a very different customer in the near future. However, randomly connecting concepts from different domains is not sufficient per se for generating something that consumers appreciate; rather, this content must also be appropriate (Rubera et al., 2010). Since foundation models have been trained on extensive datasets, it is very likely that they have memorized in the training phase what humans consider to be appropriate. Girotra et al. (2023) conducted a comparative study, pitting a pool of ideas generated by MBA students against those generated by ChatGPT-4 in two distinct conditions.

Speaking at the World Government Summit in February, he said that, given how quickly AI is advancing, the kids of today might not need coding skills to join the tech sector job market. Matt Garman, CEO at AWS, has also recently suggested that AI could mean developers aren’t writing code within the next two years. While the generative AI gold rush has sparked a surge in spending on both buying large language models (LLMs) and building their own, companies are beginning to realize that the technology is failing to live up to the promises they were sold on. By eliminating the need to define a task upfront, transformers made it practical to pre-train language models on vast amounts of raw text, allowing them to grow dramatically in size.

Given this context, consumer responses to a firm’s disclosure of GenAI use appear ambiguous. On the one hand, GenAI demonstrates superior capabilities in generating empathy and unique content that is specifically tailored to consumer needs, compared to other types of AI (Huang & Rust, 2023). Therefore, algorithm aversion should theoretically be lower for GenAI than for other AI applications.

This rapidly evolving field is changing the way we approach everything from content creation to software development, promising never-before-seen efficiency and productivity gains. Gen AI is a big step forward, but traditional advanced analytics and machine learning continue to account for the lion’s share of task optimization, and they continue to find new applications in a wide variety of sectors. Organizations undergoing digital and AI transformations would do well to keep an eye on gen AI, but not to the exclusion of other AI tools. Just because they’re not making headlines doesn’t mean they can’t be put to work to deliver increased productivity—and, ultimately, value. The first example is banking, with an estimated total value per industry of $200 billion to $340 billion, and a value potential increase of 9–15% of operating profits based on average profitability of selected industries in the 2020–22 period. The third example is pharma and medical products, with an estimated total value per industry of $60 billion–$110 billion, and a value potential increase of 15–25% of operating profits based on average profitability of selected industries in the 2020–22 period.

You can foun additiona information about ai customer service and artificial intelligence and NLP. Though initially driven by copyright concerns, our SRS framework adapts seamlessly to these new scenarios, ensuring fair revenue sharing among private data owners. Crucially, this approach addresses potential financial disagreements and facilitates decentralized model development. The utility offers a way to measure the extent to which the data sources from S𝑆Sitalic_S are responsible for generating the content. It is small if the counterfactual model is unlikely to generate the same content as the deployed model, and vice versa. Nvidia has an unassailable lead in chip technology, so unlikely to feel any competitive pressure in the medium term. Powering AI data centers has an astronomical cost, potentially adding more than 10% to peak load of the US grid system.

Hence, our adoption scenarios, which consider these factors together with the technical automation potential, provide a sense of the pace and scale at which workers’ activities could shift over time. For example, our analysis estimates generative AI could contribute roughly $310 billion in additional value for the retail industry (including auto dealerships) by boosting performance in functions such as marketing and customer interactions. By comparison, the bulk of potential value in high tech comes from generative AI’s ability to increase the speed and efficiency of software development (Exhibit 5). While we have estimated the potential direct impacts of generative AI on the R&D function, we did not attempt to estimate the technology’s potential to create entirely novel product categories.

McKinsey & Company

Businesses that reinvent with generative AI will be at the forefront of new performance standards and enhance personalized experiences. The opportunity is multifaceted and holds the potential to significantly enhance patient care. Scientists and engineers have used several approaches to create generative AI applications.

To address the copyright challenges of generative AI, we propose a framework that compensates copyright owners proportionally to their contributions to the creation of AI-generated content. The metric for contributions is quantitatively determined by leveraging the probabilistic nature of modern generative AI models and using techniques from cooperative game theory in economics. This framework enables a platform where AI developers benefit from access to high-quality training data, thus improving model performance. Meanwhile, copyright owners receive fair compensation, driving the continued provision of relevant data for generative model training.

The

trio, sometimes called “the Godfathers of AI,” shared the 2018 Turing Award for their 1980s work, their

subsequent perseverance, and their ongoing contributions. By the mid-2010s, new and diverse neural net

variants were rapidly emerging, as described in the Generative AI Models section. The landscape for neural network research thawed out in the 1980s thanks to the contributions of several

researchers, most notably Paul Werbos, whose initial work rediscovered the perceptron; Geoffrey Hinton;

Yoshua Bengio; and Yann LeCun. Their combined work demonstrated the viability of large, multilayer neural

networks and showed how such networks could learn from their right and wrong answers through credit

assignment via a backpropagation algorithm.

• Consequently, government

  policymakers around the world, and even some technology industry executives, are advocating for rapid

  adoption of AI regulations.

• Furthermore, traditional AI is usually trained using supervised learning techniques, whereas generative AI

  is trained using unsupervised learning.

• As organisations grapple with AI’s disruptive potential, the key lies in creating customer value while preparing for larger shifts.
• They then independently develop intelligence—a representative

  model of how that world works—that they use to generate novel content in response to prompts.

• Powering AI data centers has an astronomical cost, potentially adding more than 10% to peak load of the US grid system.

Specifically, for a randomly sampled permutation of copyright owners, we can first train on the first copyright owner, then the second, and all the way up to the last copyright owner. This technique can be used together with the famous permutation sampling estimator for the Shapley value [24]. From the definition of the SRS in (2.4), if a contributor’s data has a relatively large Shapley value, this contributor would receive a large royalty share, and vice versa. As the Shapley value is a fair metric of each party’s contribution to the coalition [34], the SRS offers a principled approach to assigning royalty shares to copyright owners. A related approach is called leave-one-out (LOO) score [5, 6], which examines the effect of removing a single data point or source from the entire training dataset. This shortcoming becomes especially pronounced with data duplication across various copyright owners, which is common in machine learning applications [20]; see the supplementary materials for detailed discussion.

This approach might not accurately reflect the significance of a data point due to potential complex interactions among data sources. Duplicated data points are prevalent across many widely-used machine learning datasets [20]. The removal of either from the dataset would likely result in minimal change to the model’s content generation likelihood, rendering both LOO scores close to zero. This scenario could unjustly allocate no royalty share to either contributor, despite their datasets’ crucial role in model performance. Moreover, in situations with numerous data sources, the LOO score might diminish to near zero, failing to recognize the nuanced contributions of individual sources. We validated these research questions via in-depth interviews with managers from different industries, spanning from high-end fashion and fast-moving consumer goods to insurance and utilities.

Capitalizing on Galactica’s failure when it launched ChatGPT, OpenAI explicitly acknowledged that it could make mistakes. It works better when it is tasked with generating novel content for which there are no right or wrong answers (e.g., artistic content, novel product ideas). Self-supervised training yields two direct consequences that define GenAI’s ability to generate new, plausible content. First, since the original data contains the correct part to predict (i.e., the missing part), the training process can scale to very large datasets, which would be challenging to annotate by hand.

Gen AI is used to analyze medical images and assist doctors in making diagnoses in the healthcare industry. Up to 50% of all medical errors in primary care are administrative errors, according to a report by the World Health Organization (WHO). Gen AI has the potential to increase accuracy but the technology also comes with vulnerabilities.

GPT-3, at 175 billion parameters, was the largest language model of its kind when OpenAI released it in 2020. Other massive models — Google’s PaLM (540 billion parameters) and open-access BLOOM (176 billion parameters), among others, have since joined the scene. Transformers processed words in a sentence all at once, allowing text to be processed in parallel, speeding up training.

Then AI programs, especially those with image-generating capabilities, can create detailed

designs of potential products before simulating and testing them, giving workers the tools they need to make

quick and effective adjustments throughout the R&D cycle. Generative AI has the potential to automate certain tasks, displacing some workers, and it can also create new jobs and industries. The impact of AI on jobs is difficult to predict and will likely vary depending on the industry and the specific tasks involved.

The potential improvement in writing and visuals can increase awareness and improve sales conversion rates. Our analysis captures only the direct impact generative AI might have on the productivity of customer operations. Our second lens complements the first by analyzing generative AI’s potential impact on the work activities required in some 850 occupations. We modeled scenarios to https://chat.openai.com/ estimate when generative AI could perform each of more than 2,100 “detailed work activities”—such as “communicating with others about operational plans or activities”—that make up those occupations across the world economy. This enables us to estimate how the current capabilities of generative AI could affect labor productivity across all work currently done by the global workforce.

Companies in these industries are adopting AI technology to optimize operational processes, enhance data analytics, and provide services closer to customer needs. According to forecasts, spending on AI systems in the global financial industry will grow significantly and dominate in 2023. This trend can be attributed in part to the need for large amounts of data and reports to be processed in financial businesses, and the introduction of AI technology can significantly improve the efficiency of data processing and business operations.

When that innovation seems to materialize fully formed and becomes widespread seemingly overnight, both responses can be amplified. The arrival of generative AI in the fall of 2022 was the most recent example of this phenomenon, due to its unexpectedly rapid adoption as well as the ensuing scramble among companies and consumers to deploy, integrate, and play with it. Over the years, machines have given human workers various “superpowers”; for instance, industrial-age machines enabled workers to accomplish physical tasks beyond the capabilities of their own bodies. More recently, computers have enabled knowledge workers to perform calculations that would have taken years to do manually. Notably, the potential value of using generative AI for several functions that were prominent in our previous sizing of AI use cases, including manufacturing and supply chain functions, is now much lower.5Pitchbook. This is largely explained by the nature of generative AI use cases, which exclude most of the numerical and optimization applications that were the main value drivers for previous applications of AI.

GenAI has the potential to fundamentally change the marketing function – from storyboarding to creative content to customization for different media channels and audiences. The potential economic benefits of generative AI include increased productivity, cost savings, new job creation, improved decision-making, personalization, and enhanced safety. There are also important questions about the distribution of those benefits, however, and the potential impact on workers and society. The adoption of generative AI is expected to significantly impact various industries and job markets including manufacturing, healthcare, retail, transportation, and finance.