Warning: very long, 2-part blog series. But this topic is too important to not carefully explain how we can educate and empower everyone to participate in the AI conversation. Our success as a society depends upon our ability to include everyone in this conversation.
“I love it when a plan comes together” – Hannibal Smith, “The A-Team.”
The most important blog I’ve written might be “Creating Healthy AI Utility Function: Importance of Diversity – Part I”, which highlights the importance of an inclusive, collaborative process for defining the variables and metrics that comprise a healthy AI Utility Function.
ChatGPT and its AI Generative counterparts, such as Google Bard, Microsoft Turing, and Facebook Blender, have raised everyone’s awareness of the power and dangers of Artificial Intelligence (AI). And the exploration and application of AI will grow exponentially over the next few years as more companies jump on the AI bandwagon, despite efforts by some to slow or even stop its development. But it’s too late. The genie is out of the bottle, and the race to exploit the powers of AI has already begun – whether we like it or not!
For AI to reach its full economic and societal potential will require everyone’s active participation in the meaningful, relevant, and responsible design, application, and management of AI. And I mean E-V-E-R-Y-O-N-E! It cannot just be the high priesthood of AI specialists, three-digit government agencies, and Silicon Valley tech companies.
So how do we prepare everyone to participate in the age of AI? The “Thinking Like a Data Scientist” (TLADS) methodology has been honed over many years of application, learning, and adapting to precisely do that. The TLADS methodology is an inclusive, collaborative process for defining the variables and metrics against which organizations define and measure their value creation effectiveness. And with the mandate to create healthy AI Utility Functions that guide AI models to deliver relevant, meaningful, responsible, and ethical outcomes, the TLADS methodology’s day has come (Figure 1)!
Figure 1: The Art of Thinking Like a Data Scientist (Version 2.0)
Leveraging the TLADS methodology, we can create a playbook that anyone can follow that lays out their role in the definition, design, development, deployment, and management of AI models that deliver relevant, meaningful, responsible, and ethical outcomes.
Step 1) Defining Value
The key to developing effective, relevant, ethical AI models is building a healthy AI Utility Function comprised of a diverse set of variables and metrics that guide the AI model’s decisions, actions, and ongoing learning and adapting.
As those of you who know me and have followed me for several years (decades?) know, I begin my customer conversations and classroom lectures by asking, exploring, and validating how organizations create value and measure their value-creation effectiveness. That is the foundation for my “Thinking Like a Data Scientist” methodology which has been honed and enhanced through many years of customer engagements and classroom exercises.
Step 1 of the TLADS methodology guides your organization through an extensive, collaborative exercise in identifying how your organization creates value, the desired outcomes from those value-creation processes, and the KPIs and metrics against which outcomes’ effectiveness will be measured. Then Step 2 expands and validates that value definition and measurement exercise by looking at the organization’s value creation processes through the eyes of its internal and external stakeholders – identifying their desired outcomes and the KPIs and metrics against which these stakeholders measure the effectiveness of those desired outcomes (Figure 2).
Figure 2: Template 1 of “The Art of Thinking Like a Data Scientist” Methodology
Value can be a very subjective definition, which is why organizations need to engage a diverse set of internal and external stakeholders whose differences in perspectives are critical in developing AI models that serve everyone.
Step 2) Understanding How AI Works
Artificial Intelligence is a set of algorithms that seek to optimize actions by mapping probabilistic outcomes to utility value within a constantly changing environment…with minimal human intervention
AI is a simple tool (not a living creature) that does exactly what you train it to do. The AI model seeks to optimize its actions to achieve desired outcomes as framed by user intent. The AI model constantly interacts with its environment and evaluates input variables and metrics and their relative weights to assign a utility value to specific actions (as defined by its AI Utility Function). Finally, the AI model measures the effectiveness of its actions so that it can continuously learn and update the weights and utility values that comprise the AI Utility Function…with minimal human intervention (Figure 3).
Figure 3: What is Artificial Intelligence (AI)?
Yea, pretty straightforward. But there is a lot of power in successfully orchestrating those five capabilities, but at a high level, that is all that AI does. Invest the time to understand AI as a tool (and leave the AI boogie man to Hollywood and the tinfoil hat-wearing conspiracy mongers hanging on social media and pontificating on your favorite non-news news channel).
Step 3) Understanding the Role of the AI Utility Function
AI Utility Function consists of variables, metrics, and associated weights that guide the AI model’s decisions, map desired outcomes to utility values, and measure decision effectiveness to continuously learn and adapt.
The AI Utility Function is the key to the effective, meaningful, and responsible execution of your AI models. The AI Utility Function assigns values to the AI system’s actions based on the economic value of outcomes. The AI models’ preferences over possible outcomes can be captured by a function that maps these outcomes to a utility value; the higher the number, the more that agent likes that outcome (Figure 4).
Figure 4: AI Utility Function
AI is like the children’s game of Hotter-Colder, with the AI Utility Function guiding the AI model’s actions in seeking the prize or desired outcome. The only difference? The prize or desired outcome is constantly moving or changing in the real world.
Summary: AI for Everyone – Part 1
For AI to reach its full economic and societal potential will require everyone’s participation in the design, application, and management of relevant, meaningful, and ethical AI models. To accomplish that, we must educate and empower everyone to think like a data scientist.
The Thinking Like a Data Scientist methodology is an inclusive, collaborative process for defining the variables and metrics against which organizations define and measure their value creation effectiveness. The TLADS methodology can help us create healthy AI Utility Functions that guide AI models to deliver relevant, meaningful, responsible, and ethical outcomes.
We can leverage the Thinking Like a Data Scientist methodology to create a playbook that anyone can follow to design, develop and deliver responsible and ethical AI models including:
- Step 1) Defining Value
- Step 2) Understanding How AI Works
- Step 3) Understanding the Role of the AI Utility Function
In part 2, we will continue the AI for Everyone playbook to ensure that everyone understands their role – and their responsibility – in developing responsible and ethical AI models.