At this stage of your process, you have covered evaluation, selection and procurement of AI technology. With your technology purchased, you’re ready for launch, but before, or at least during, the roll-out, it’s critical to create the proper cultural and structural framework in which the AI will operate.
This matters. AI is a complicated technology. It is loaded with all types of connotation and innuendo, and is especially challenging to monitor, measure and interpret. Establishing a framework for your AI deployment will determine success for your business and the people using it.
Address Cultural Change
There’s an unbelievable amount of content out there on AI, chatbots, facial recognition, driverless cars, and tons of other applications, lots of which is negative. There will likely be some level of resistance, or at least skepticism, of the technology. However, AI is a tool like any other. It has a structure, operating mechanics, necessary conditions to operate, and ideally, a solution it delivers. Onboarding new tools requires change.
For the cultural onboarding, consider taking the follow steps:
Develop a change management plan. To be successful, implement a step change approach, complete with milestones, education session and initiatives that bring people onboard. Because AI often depends on people interacting with the algorithm, you may consider some types of incentives for power users.
Education sessions. These should be geared towards communicating the technology and the goals of the project, as well as the positive impact the users can expect. You should focus on the mechanics of AI with an aim to demystify the ins and outs of the technology and how it works. Creating space between human and robotic work will make this time well spent.
Critical roundtable. If education sessions are factual, subsequent roundtables should engender an honest dialog. People will have justified concerns about AI, which should be addressed head on, and will likely have questions about the technology and the future of their roles. Get these discussions going at full steam, so there's fewer doubts going into the project, and will perhaps help roll out for other parts of the business.
Brainstorm job evolution. While you can prescribe what the team should focus on in a post-AI world, it’s more valuable to take this theme directly to the team. Get their views on transformation and the type of work they consider valuable. They will be more bought into activities like increased personalization, or upselling and you might discover some previously unconsidered positive externalities that AI can bring about.
Testing. Provide opportunities for the team to see it in action or interact with the technology, which will dispel nagging anxiety, perhaps stimulate new ideas. This is also a useful socializing step for the technology since it should highlight how AI is not always right, and human input is required for success.
Creating a Structural Framework
The other side of the coin is the structural conditions for success. If you got this far, well done, you've convinced your business to invest time and resources in an AI project, but what's next? Few, if any businesses, are purpose built to operate AI. There are policies to consider, information to share and organizational hierarchies to shift.
Consider these three elements in adapting your business framework:
Collaboration interface. How will everyone work together, what do they need to get started, and to maintain or improve the technology? Users, developers, testers and data scientists must have a framework in which to operate and collaborate. They need tools, training and incentives to ensure the AI is constantly learning and evolving.
Decide who owns it. Too many times these projects get stuck because there’s no clear owner with final say. Is it the data science team? Or will it be delegated to business units? Or perhaps an enterprise-wide AI leader needs to be created to have a seat at the management table? Either way, final say and accountability is a critical part of your structure to ensure the project is properly enabled.
Fairness & accountability. You must assign responsibility for fairness, interpretability and security of AI, with clear roles and expectations for all parties. In this way you will build trust in the product and the viability in the outcome and the longer term project.
There are many steps to getting going with AI: technical onboarding, discovery, integration, data training and model creation, and numerous others, however, planning the cultural and structural onboarding is just as critical.
The framework you set up will define the context within which the AI project will operate and hopefully succeed. As with any project, you need people and process to function together in support of the project goals.