Advantage Digital: How can IT ready itself for AI? | KPMG | UK
close
Share with your friends

How can IT ready itself for AI?

Advantage Digital: How can IT ready itself for AI?

Leading the way on the AI journey are IT departments, but what do they need to ensure they succeed in implementing this change?

1000

Also on KPMG.com

How can IT ready itself for AI? - robot hand holding a lightbulb

The successful businesses of the future will be those that have placed artificial intelligence at the very heart of what they do.

New technologies, including AI and one of its most exciting applications, machine learning, are transforming how organisations operate at a pace that will only accelerate.

Leading the way on this AI journey are IT departments, but what do they need to ensure they succeed in implementing this momentous change?

1. Build a solid foundation

One of the main obstacles to applying machine learning is that organisations underestimate what is required in terms of resources and commitment.

This can partly be solved through strong leadership in the central data team and by having a senior advocate, preferably at board level. The goals for machine learning need to be aligned with business priorities from the start, rather than just implementing technology because it is available.

Companies also need a strategy for how machine learning teams will engage with the rest of the business and set a clear pipeline of priorities. Otherwise they can be bogged down with distracting or overwhelming demands.


2. Finding the right talent

Finding the right talent is not simply a case of hiring Einsteins. Being clever is not enough, they have to be able to communicate in a way that encourages business leaders to be on board with the technology transformation and the AI journey.

However, if organisations want to attract the best talent they may need to break salary bands to be competitive, and they must offer clear training and progression opportunities for team members.

It is also important to establish the right tactical environment early on, such as freeing the team from IT obstacles like having to obtain local admin rights and join permission groups.

3. Establishing machine learning capability and data science

During the early stages, companies need to focus on easily achievable projects, like those requiring minimum dependencies and data feeds, so there are fewer variables to go wrong.

Early wins can be used as marketing collateral and will ensure continued support for future projects.

While having robust data and strong data scientists is a prerequisite, it is also useful to have a “data lake” – where unstructured data can be stored for use later – without worrying about how it is going to be processed or when it will used.

4. A change in culture

A successful machine learning journey will often necessitate a transformation in company culture.

There has to be a willingness to experiment, even on the part of customers, as well as an acceptance that things might not always go as planned. This is fundamentally a scientific process, and organisations that accept this and learn quickly from any failures are likely to be the most successful.

Making sure everyone in the organisation is data literate and understands its importance is also essential.

5. Governance

Machine learning throws up ethical, governance and compliance issues, such as how we use consumer data and how a supply chain interacts with that information.

Companies also must consider what happens when an algorithm makes a mistake. Who takes the blame when someone is not offered a financial product or has received unequal treatment?

Data scientists should expect to be held accountable. Consequently, they must make sure that an algorithm is penetrable; that a simple explanation of how it reached a decision can be easily found.

There needs to be a consensus of good practice within the industry, and professionals must work with regulators and governments in writing policy or have regulations imposed on them.

Five to 10 years from now?

The real question is not what will be different five or 10 years from now, but what will not be different?

Undoubtedly, the languages of AI and machine learning will be more commonly used. We will get to the stage where society and business will not worry about how to engage with them as they will be the background of all we do.

To find out more about how digital transformation can impact your organisation, contact us.

Connect with us

 

Request for proposal

 

Submit