A third of AI projects will be abandoned by businesses by the end of next year

Gartner has predicted that by the end of next year, 30% of generative AI projects initiated by businesses will be abandoned after their proof of concept stages. The reasons for this anticipated abandonment include poor data quality, insufficient risk management, rising costs, and unclear business value. Gartner highlighted various applications of generative AI, such as coding assistants, personalized sales content creation, document search using Retrieval-Augmented Generation (RAG), virtual assistants, and large language models (LLMs) for sectors like medical, insurance, and finance. Coding assistants are noted as the most cost-effective, with initial costs between $100,000 and $200,000 and annual user costs ranging from $280 to $550. In contrast, developing LLMs for specialized services can incur upfront costs of $8 million to $20 million and annual costs of $11,000 to $21,000. Rita Sallam, a Distinguished VP Analyst at Gartner, commented on the impatience of executives for returns on generative AI investments, emphasizing the challenges organizations face in demonstrating value. The financial burden of developing and deploying generative AI models is becoming increasingly significant, with costs varying widely based on use cases and deployment strategies. While some businesses may discontinue certain generative AI applications, the evolving capabilities of language models may prompt them to explore new use cases in the future.