​​​​​​​​​The figure above shows the GenAI timeline. The year 2023 can be considered as the Year of Discovery, year 2024 as the Year of Piloting, and years 2025 and beyond as the Years of Adoption.

​

As per a study by Boston Consulting Group, 10% of firms have scaled generative AI, while 90% of them are still lagging. Out of these 90%, 50% have begun piloting, where as 40% have taken no action. 

​

Out of 50% of those who have begun piloting, a staggering 95% of them have committed the following three piloting mistakes:

​

Mistake 1: Random selection of pilots

The pilots were selected randomly without a scaled adoption plan in place. These pilots are called nonconvertible pilots. The convertible pilots are those that have higher probabilities of moving into production and eventually scaling.

​

There are three outcomes possible for a pilot:

​

1. It fails

2. It is successful but fails to scale

3. It is successful and scales successfully

​

A nonconvertible pilot has 1. and 2. as its possible outcomes.

A convertible pilot can have all the above three outcomes possible. It can fail too. But the chances of success are higher.

​

​

Mistake 2: No cross-functional alignment

The pilots were conducted within functions in silos, without much alignment with other functions. For example, sales function had their own pilots, which was different from finance or supply. There was no visibility of pilots in other functions.

There was also little involvement or support from the leadership team.

​

Mistake 3: Standalone GenAI

For scaling, genAI needs to be deployed in a collaborative mode with adjacent technologies such as traditional AI, data, cloud, and automation. The scope for most of the pilots only included deploying genAI in a standalone mode, completely overlooking other technologies. As a result, these pilots are likely to face even greater challenges when transitioning to production, as the deployment teams will lack the knowledge, experience, and collaborative strategies needed to integrate them effectively with other adjacent technologies.

 

​

Lessons learnt:​

1) Select convertible pilots

​

2) Align with cross functional teams and secure leadership support

​

3) Deploy genAI in a collaborative mode.

​

​

​

​

​