Just Enough AI Governance

Artificial intelligence continues to be top of mind for global leaders as they came together for the G7 last week.? In their final statement, leaders noted:

"We will pursue an inclusive, human-centred, digital transformation that underpins economic growth and sustainable development, maximizes benefits, and manages risks, in line with our shared democratic values and respect for human rights.”

As larger organizations look to harness the opportunities of Digital Transformation through the application of AI, many have launched efforts to create governance structures to focus the oversight of their development of AI enabled tools and business processes.? In part two of this series, I mentioned these efforts and the assignment of accountability to Chief Privacy Officers and Chief Data Officers.? But what’s a smaller organization to do?

Smaller organizations won’t have a Chief Data Officer, or a Chief Privacy Officer. For that matter they might not have a Chief Anything Officer, so accountability falls on the business owner.? Let’s think about what that accountability might look like.? For all those small businesses that won’t be building the next-generation AI models, AI Algorithms or AI enabled applications, those small business owners will be leveraging practical tools that have been AI enabled behind the scenes.? AI enabled applications like voice to text, or AI powered service manuals or AI enabled fundraising software, or…the list goes on.? I would argue that these “behind the scenes” uses of AI fall outside an AI governance framework.? The federal government’s Guide on the use of generative artificial intelligence I mentioned in the last article call out these types of AI solutions that don’t make administrative decisions as being outside their Directive on Automated Decision making and assert that “using generative AI tools in research or to brainstorm, plan, or draft routine correspondence falls outside the scope of the directive.”? Also to note is that the UK Guidance - Generative AI Framework for HMG calls out “Public Gen AI” and “Embedded Gen AI Applications” as distinct from “APIs” and “local development.” Most SMEs will leverage the public generative AI solutions or embedded AI solutions where there is little customization to address responsible AI principles (the service providers do this in the shared responsibility model – See Microsoft Responsible AI for example ).? As a result, most small enterprises should simply look to becoming skilled with the capabilities and pitfalls of use of generative AI and not let the burden of a heavy governance process prevent their adoption.? There are many materials online to help organizations skill up with simple to use guidance like this from the UK

: