Despite the promise that artificial intelligence will transform industries, rising costs and increasing risks are causing many AI projects to fail, as highlighted by several recent reports.
According to a new report from Gartner, at least 30% of generative AI projects will be abandoned after the proof-of-concept stage by the end of 2025. Companies are “struggling to demonstrate and realize value” in their initiatives, which cost between $5 million and $20 million in initial investments.
A separate report from Deloitte found a similar result. Of the 2,770 companies surveyed, 70% said they had only moved 30% or less of their GenAI experiments into production. Lack of preparedness and data-related issues were attributed to this low success rate.
The overall outlook for AI projects is not encouraging. Research by the RAND think tank found that despite private sector investments in AI increasing 18-fold between 2013 and 2022, more than 80% of AI projects fail – twice the failure rate for corporate IT projects that do not involve AI.
The disparity in financial backing and completion is a likely contributor to the “Magnificent Seven” tech companies — NVIDIA, Meta, Alphabet, Microsoft, Amazon, Tesla and Apple — losing a combined $1.3 trillion in stock value over five days last month.
SEE: Nearly 1 in 10 companies will spend more than $25 million on AI initiatives by 2024, according to a report from Searce
High initial investments are required in GenAI projects before benefits are realized
According to Gartner, using a GenAI API (an interface that allows developers to integrate GenAI models into their applications) could cost up to $200,000 upfront and an additional $550 per user per year. Additionally, creating or fine-tuning a custom model can cost between $5 million and $20 million, plus $8,000 to $21,000 per user per year.
According to a report by automation software provider ABBYY, the average investment in AI by global IT leaders was $879,000 last year. Almost all (96%) of respondents said they would increase these investments in the coming year, although a third said they are concerned about the high costs.
Gartner analysts wrote that GenAI “requires a higher tolerance for indirect and future financial investment criteria versus immediate return on investment,” which “many CFOs have been uncomfortable with.”
But it’s not just CFOs who have doubts about the return on investment from AI projects. Investors in the world’s biggest tech companies have recently expressed doubts about when, or if, their support will pay off. Jim Covello, an equity analyst at Goldman Sachs, wrote in a June report: “Despite its high price tag, the technology is a long way from reaching the level needed to be useful.”
WATCH: UK tech startups suffer first decline since 2022, down 11% this quarter
Additionally, the market values of Alphabet and Google declined in August as their revenues failed to offset their investments in AI infrastructure.
Other causes of the failure of the GenAI project
A key reason for failure in launching enterprise GenAI projects? Lack of preparation.
Less than half of Deloitte respondents felt their organizations were well prepared in the areas of technology infrastructure and data management, two building blocks needed to scale AI projects to a level where benefits can be realized. The RAND study also found that organizations often do not have the “appropriate infrastructure to manage their data and deploy full AI models.”
Only about 1 in 5 Deloitte respondents indicated they were prepared in the areas of “talent” and “risk and governance,” and as a result, many organizations are actively hiring or upskilling for AI ethics-related roles.
SEE: 83% of UK companies increase salaries for professionals with AI skills
Data quality represents an additional obstacle to completing GenAI projects.
Deloitte’s study found that 55% of companies have avoided certain GenAI use cases due to data-related concerns, such as data sensitivity or privacy and security concerns. RAND’s research also highlighted that many organizations do not have the data necessary to train an effective model.
Through interviews with 65 data scientists and engineers, RAND analysts found that the number one reason an AI project fails is a lack of clarity about the problem it promises to solve. Industry players often misunderstand or miscommunicate this problem, or choose one that is too complicated to solve with the technology. The organization may also be more focused on employing the “latest and greatest technology” than actually solving the problem at hand.
Other concerns that may contribute to the failure of the GenAI project cited by Deloitte include the inherent risk of AI (hallucinations, bias, privacy concerns) and keeping up with new regulations such as the EU AI Act.
Companies remain steadfast in their search for new GenAI projects
Despite low success rates, 66% of US-based CIOs are in the process of implementing GenAI copilots, up from 32% in December, according to a Bloomberg report. The top use case cited was chatbot agents, such as those in customer service applications.
The percentage of respondents who said they were currently training role models also increased from 26% to 40% in the same period.
The RAND report provided evidence that companies were not reducing their GenAI efforts as a result of difficulties in implementing them. According to a survey, 58% of midsize companies have already deployed at least one AI model in production.
According to Gartner, this persistence in GenAI is due to some tangible impacts on revenue savings and productivity. Meanwhile, two-thirds of organizations surveyed by Deloitte said they are increasing their investments because they have seen strong initial value.
However, ABBYY research found that 63% of global IT leaders are concerned that their company will fall behind if they don’t use it.
There’s even evidence that GenAI is becoming a distraction. According to IBM, 47% of tech leaders feel their company’s IT function is effective at delivering core services, a 22% decrease since 2013. Researchers suggest this is related to the fact that they are turning their attention to GenAI, as 43% of tech executives say they have increased concerns about infrastructure in the past six months.
Rita Sallam, vice president analyst at Gartner, said: “These data serve as a valuable benchmark for assessing the business value derived from GenAI business model innovation.
“But it’s important to recognize the challenges involved in estimating that value, as the benefits are very specific to each company, use case, function, and workforce. Often, the impact may not be immediately apparent and may materialize over time. However, this delay does not diminish the potential benefits.”
