AI Implementation

A robot sketch drawn for an article about chatbots. by James Royal-Lawson

Whose job is it to implement AI/ML into an organization? This week we ask a hard question and propose a solution that may result in AI/ML creating jobs…

A question to ponder
The Feb 2018 Deloitte CMO survey asks participants a simple question: 

To what extent is your company implementing artificial intelligence or machine learning into its marketing toolkit?

It’s not a surprise that a lot of answers come in the “Not at all range” and I have been using this example to show AI/ML has low implementation within organizations. But recently I have been thinking about this question in a different way, leading to another question:

Who is responsible for implementing artificial intelligence or machine learning into your company?

It would be very interesting if you asked everyone in an organization that question and see what the results are. To be honest, I don’t have a good answer to the question. Originally, my opinion was that the analytics/data science function should be responsible for bringing AI/ML, as these groups already use the math and programming that underpins AI/ML, why not they just do it? 

My opinion is naive
Of course, the issue is much more complicated than that. As the application of AI expands from building predictive models, to things like robotic process automation, robots, image recognition, or deciding how new advances in AI should be applied, there suddenly becomes too many things analytics/data scientists can look after. 

There is also the issue of domain experience. Every domain from marketing, finance, HR, etc. will benefit from AI/ML, but that does not mean that you can apply the same solution to each business function. Each has nuances that only specialized professionals know about and this information needs to incorporated into AI/ML. In this scenario, should the C-Level execs of marketing, finance and HR be the ones responsible for implementing AI/ML? 

What’s the ideal solution?
The most ideal solution would be to have the data science/analytics functions working with the business line executives to implement the technology. This will work for basic implementations, but as the technology advances, I don’t think this solution will work. 

Ultimately, I believe large organizations will have to create an internal AI division, which would be a combination of machine learning engineers, business analysts and product managers. These people would be responsible for driving adoption, building solutions, managing hardware, and staying on top of the latest advances. The responsibilities of this group could also include ethical implementation of AI and how to manage automation, but that’s asking too much for a group we just made up.

If AI/ML hasn’t been integrated into a business function, the groups first project would be building an AI/ML strategy, which would include organizing data resources, identifying business problems and creating a plan of action. After, the group would be responsible for either building the solution or partnering with a 3rd party to build it. If a business function has integrated AI/ML, the AI groups function could be measuring the success of the initiative and keeping the business function on top of any advances. 

Ideally, the AI group would sit under the CTO, as after all AI is a technology. I am not advocating for a chief AI officer, as others have proposed, as we still need to see how quickly AI/ML takes off. If only 25% of companies have a chief digital officer, after we all know digital is the overwhelming future, then creating a chief AI officer will take some time. 

As always…
I could be wrong and this whole AI thing ends up being a flop, causing layoffs within the AI division, but then again, AI was always expected to kill jobs…

Facebook Comments