Overcoming AI bias through cross-functional innovation

The Ether with Casey Simple: Cross-Functional AI Innovation

Add bookmark

Seth Adler

Sponsored Content

Casey Simple on Cross-Functional AI Innovation

I had a chat with a millennial data analyst who's working in an AI company, who had read The Ether and posed a couple of good questions. The first was; How does my team know what to raise as an issue if we are ethically compromised? And the second was; How do we know if it's an ethics issue or an organizational culture issue?

Enterprise-wide AI

Both of these questions I think lead us to an important principle for successful AI innovation. It's a little roundabout way to get to the topic, but generally teams are ethically compromised due to some unconscious bias. That's why it's really important to have AI teams that are not singularly focused and that are not siloed from the operational areas of the company. By making AI an enterprise-wide activity as opposed to a siloed activity, you then end up creating a cross functional focus which can overcome that single team bias.

That understanding brings us to the second question on organizational culture. If the organization drives to expected results with inflexible deadlines without allowing for the unexpected- there's definitely a cultural issue. But there's also a structural issue that is more easily overcome. If the organization is set up in silos with one group designing, another developing, a different team implementing and no collaboration at any of those stages- you run the risk of missing innovation opportunities because you've siloed those teams.

Learning from past evolution

We have a great example of how to overcome this kind of siloed approach. If we look back in time at the operational areas, as we began to develop shared services and its evolution over the past couple of decades.

As shared service centers grew from singularly focused teams that were just operating as ways to increase operational efficiency and began to grow into multifunction, multi locations centers of operational expertise- they also became hotbeds of innovation in operational processes and became the leaders in those kinds of innovation areas.

My suggestion is that the same could be done with AI development and implementation. You want to engage and involve the operational teams early and often. So if you have a shared service center, you can make them a part of your AI design and implementation teams. Thus additional cross functional collaboration will begin to ignite innovation and keep your AI teams from becoming biased and singularly focused. It'll also give them exposure to real operational issues with their AI development ideas. And it keeps teams from unconscious bias as they are exposed to other ways of thinking. There's no reason that a multifunctional cross collaboration structure for AI design, development and implementation, just like we've had with operational areas, can't become the newest hotbed for AI innovation.