Nearly all businesses are saying GenAI initiatives are failing, according to a new report from Massachusetts Institute of Technology (MIT).
The GenAI Divide: State of AI in Business 2025 report, published by MIT’s NANDA initiative, found that 95 percent of businesses that have implemented GenAI initiatives to build efficiency and drive growth aren’t delivering.
The report surveyed 350 employees, carried out 150 interviews with business leaders, and analysed 300 public AI deployments. From this, it found that only five percent of AI pilot programmes achieved growth in-line with desired outcomes.
Integration and Adoption
The main causes cited in the report include a ‘learning gap’ in what organisations understand about AI and the tools themselves, and failed integration of AI tools into workflows.
For those companies that saw some success – empowering managers to drive adoption and selecting AI tools that can easily be adopted and adapted over time, were seen as important.
Another reason businesses haven’t seen desired outcomes from their AI efforts is misallocation of the tools.
MIT’s research concludes that companies are spending over half their AI budgets on marketing and sales, when the best return-on-investment comes from AI replacing back-office automation.
Internal Models vs External Partnerships
Partnering with AI providers as opposed to building your own internal tools is more likely to end well – with external partnerships succeeding 67 percent of the time compared with one-third of the time.
Measuring the success of AI tool integration continues to be a challenge for most businesses, the research found.
