Science & Tech · AI · 1 hr ago
Real-world examples helped companies find more ways to use AI
Many companies have adopted AI but have not turned that investment into clear gains in productivity or performance.
Researchers call the challenge finding where AI can change work and which processes need to be redesigned.
In a field experiment, 515 early-stage companies in several countries received AI tools, technical training, mentoring and credits to use AI services.
Half also saw examples of how other businesses redesigned their work with AI, then were encouraged to rethink their own processes.
Those companies identified 44% more AI uses, completed 12% more tasks and were 18% more likely to win paying customers than the comparison group.
They generated nearly twice as much revenue and needed significantly less outside capital to grow.
The researchers say companies should map work from start to finish, look for delays and handoffs, and expect new bottlenecks when AI speeds up one part of a process.
Companies shown real-world examples of how others used AI identified 44% more AI use cases than companies without the examples.
In a field experiment involving 515 early-stage companies in various countries, all participants received advanced AI tools, technical training, mentoring and API credits.
Companies exposed to the examples completed 12% more tasks, were 18% more likely to gain paying customers and generated nearly twice as much revenue.
The research says examples helped managers see more ways to redesign workflows and business models with AI.
Researchers describe the difficulty of identifying where AI can change a business as the “mapping problem.”
- Who
- Researchers Hyunjin Kim and Dahyeon Kim of INSEAD and Rembrand Koning of Harvard Business School studied 515 early-stage companies.
- What
- A field experiment found that companies shown real-world AI examples identified more use cases and achieved stronger results.
- When
- The research is described as taking place nearly four years after the spread of large language models. The experiment date is not stated.
- Where
- The experiment involved companies in various countries.
- Why
- The researchers examined why many companies struggle to identify where and how AI can transform their businesses.
This story does not have two clearly opposing sides.
Many companies had adopted AI but had not turned that investment into productivity or performance gains.
Researchers gave 515 early-stage companies AI tools, training, mentoring and API credits; half also saw examples of AI-driven workflow and business-model redesign.
Companies shown examples identified 44% more AI use cases than the control group.
- Companies in experiment
- 515 early-stage firms
- Use cases identified
- 44% more in the group shown real-world examples
- Tasks completed
- 12% more by companies shown examples
- Paying customers
- 18% greater likelihood among companies shown examples
- Revenue
- Nearly twice as much for companies shown examples
What people are saying
What prevents the company from growing ten times more with its current structure?