The global market for generative AI will grow from $6bn this year to $59bn in 2028, new forecasts show, as the technology is deployed in industries ranging from marketing to insurance.
Few software markets, if any, have grown as fast or produced as much excitement — and fear — as generative artificial intelligence. Supporters predict that the technology — which can generate vast amounts of sophisticated media, such as text, computer code, images or music by learning from online data — will facilitate huge increases in productivity. Others warn about the technology’s potential to do harm — including automating hundreds of millions of jobs or even posing a threat to humanity.
“Generative AI is innovative and disruptive,” says Eden Zoller, chief analyst at Omdia, which has researched the future of the generative AI software market. “Many businesses are experimenting with it, and some are getting good results. The technology has huge potential but also huge risks.”
Growth will be fastest in the consumer market — where AI can power search engines, writing assistants, and artwork production — with revenues forecast to hit about $11bn by 2028, Omdia says. Next will come the media and entertainment sector, which will hit $8bn in revenue, followed by the healthcare, business, and automotive markets.
Generative AI businesses can broadly be divided into three categories: creators of huge “foundation models” trained on massive amounts of data scraped from the internet; providers of the necessary infrastructure, such as computing power; and providers of software that adapts the technology to business tasks or industries.
Those first two categories are dominated by a few companies, because of the scale of the computing power and knowhow required. Key suppliers in the foundation model market include OpenAI, Google and Anthropic, a start-up. In infrastructure, the main suppliers include Amazon Web Services and Nvidia — whose powerful graphics processing unit chips help to meet AI’s intense computing demands.
But in the third category — creators of products that run off generative AI — there is a much wider range of companies.
One example, in the marketing sector, is Texas-based company Jasper. Its tool automates tasks such as creating marketing campaigns and writing social media posts. The company — which has clients including HubSpot, a customer relationship management platform, and the magazine Sports Illustrated — has adapted models from companies including Anthropic, Google and OpenAI.
Jasper uses the technology to deliver efficiencies. One if its other clients, a big carmaker, wanted to develop a new marketing campaign for electric vehicles and was initially told that it would take two weeks. “Jasper was able to automate that workflow in two days,” says Suhail Nimji, Jasper’s head of business development.
However, the company does not believe its services will cause employers to fire all their marketing staff and replace them with generative AI. “It’s not [about] reducing or removing humans from the cycle, it’s about optimising [their work],” argues Nimji. “Jasper writes the first [draft of a] blog and then a human takes over.”
Generative AI is also finding a market in legal services and insurance by helping to automate laborious but important tasks — such as gathering, analysing and creating documentation. This can include drafting large commercial contracts and spotting risks, such as irregularities in documents.
For example, law firm Womble Bond Dickinson is using generative AI software tools for brainstorming ideas and drafting documents. One such tool is Lega, which helps law firms and other businesses “safely explore . . . and implement” generative AI technologies.
Chris Mammen, a partner at Womble Bond Dickson, says that early results of the software, which it has been using since the start of 2023, are encouraging. Generative AI will help it “deliver our client services, better, faster and cheaper”, he says.
Yet emerging flaws in generative AI — such as making up facts, which is referred to as “hallucinating” — means careful oversight is needed. Mammen stresses that lawyers always review the software’s output.
The firm is also building its own internal chatbot to assist lawyers. This chatbot is trained on the firm’s policies and procedures — enabling it, for example, to answer a lawyer’s questions about a type of expense claim by generating correct answers extracted from all relevant policies.
Insurance technology supplier Cytora is also employing AI. It says that one customer, the speciality insurer Markel, improved the output of underwriters by 113 per cent, by using its software to automate much of the data analysis and risk assessment involved in writing an insurance policy.
“Insurance companies are bombarded with . . . information and all of it is different,” explains Richard Hartley, Cytora chief executive. “Generative AI, at extremely low cost, can read all of this, digitise it, standardise it and highlight areas that are unusual.”
Consumer uses for generative AI are growing rapidly, too. New applications include a forthcoming tool in messaging app WhatsApp that can instantly draw cartoons to order. WhatsApp says it can attempt to depict anything, even a raccoon driving a motorcycle.
Meta, which owns WhatsApp as well as Facebook and Instagram, is also developing an assistant that can sit within a group chat so, if there is a discussion about going on holiday, for example, it can present information on destinations gathered from the internet.
Last week, Microsoft announced it would also include a generative AI-powered assistant, called Copilot, in its Office software, for an extra $30 per month per user. The feature would be able to draw on data held in a company’s existing Word documents or Excel spreadsheets.
Some software companies will inevitably suffer, though. As the generative AI software market matures, smaller suppliers are likely to disappear — as was the case in the automotive market in the 1930s, says Euan Cameron, head of artificial intelligence in the UK for PwC, a professional services firm.
“The generative AI market probably won’t continue to sustain 400 different [AI] players and will see consolidation,” he says.
Read the full article here