Delivering the best member and customer experiences with a focus on trust is core to everything that we do at LinkedIn. As we continue to build on our Responsible AI program that we recently outlined three months ago, a key part of our work is designing products that provide the right protections, mitigate unintended consequences, and ultimately better serve our members, customers, and society.

Better AI Systems

Transparency means that AI system behaviour and its related components are understandable, explainable, and interpretable. The goal is that end-users of AI—such as LinkedIn employees, customers, and members—can use these insights to understand these systems, suggest improvements, and identify potential problems (should they arise).Developing large-scale AI-based systems that are fair and equitable or that protect the users may not be possible if our systems are opaque.

The principle of Transparency

Transparency allows modellers, developers, and technical auditors to understand how an AI system works, including how a model is trained and evaluated, what its decision boundaries are, what inputs go into the model, and finally, why it made a specific prediction. This is often also described as “interpretability” in existing research. Explainable AI (XAI) goes a step further by explaining how a system works or why a particular recommendation was made to members and customers.

Transparency + Context = Explainability

A few key ways we've improved transparency in AI at LinkedIn:<ul><li>Explainable AI for model consumers to build trust and augment decision-making.</li><li>Explainable AI for modellers to perform model debugging and improvement.</li><li>Transparency beyond AI systems.</li></ul>

Improving AI Transparency

Complex predictive machine learning models often lack transparency, resulting in low trust from these teams despite having high predictive performance. While many model interpretation approaches such as SHAP and LIME return top important features to help interpret model predictions, these top features may not be well-organized or intuitive to these teams.To deal with this challenge, we developed Intellige, a customer-facing model explainer that creates digestible interpretations and insights reflecting the rationale behind model predictions.

Augmenting trust and decision-making

Machine learning engineers at LinkedIn need to understand how their models are making the decisions to identify blindspots and, thereby, opportunities for improvement. For this, we have explainability tools that allow model developers to derive insights and characteristics about their model at a finer granularity.

Explainable AI(XAI) for modellers

From a holistic “responsible design” perspective, there are many non-AI initiatives that help increase the transparency of our products and experiences.<ul><li>On the backend, researchers have previously highlighted the importance of dataset documentation as a key enabler of transparency.</li><li>On the front end, we launched a transparency initiative designed to earn and preserve member trust through improvements to our reporting and policy enforcement experiences.</li></ul>

Transparency beyond AI systems

Stashed ideas

Our approach to building transparent and explainable AI systems

The traditional approach to goal-setting is measured by one thing: the outcome. This goal-setting approach is problematic for three reasons:<ol><li>It attaches our self-worth to the outcome of our goal: If we accomplish our goal, we see ourselves as successful; if we don’t, we see ourselves as failures.</li><li>It cultivates a harmful “if-then” happiness contingency mindset that removes you from the present and fixates you onto a time in the future. </li><li>It provides no system for measuring progress as you work toward your goal, thus falsely packaging the outcome as the sole measurement of progress </li></ol>

The Traditional Approach: Outcome-Based Goals

The GAP (growth, awareness, progress) approach focuses on three pillars:<ol><li>Growth by Consistency (your inputs). Growth happens through a series of small and consistent incremental steps, not monumental overnight strides. </li><li>Awareness by Metacognition (your outputs). The objective is to stay aware of your output and reflect on whether you’re implementing the most effective strategies that’ll help you achieve your goals. This places extra emphasis on the process and your learning loops.</li><li>Progress by Tracking (your logs). This third pillar is an extension of the first two. By tracking your inputs (placing an X on your </li></ol><a class="idea-content-link" href="https://www.omaritani.com/habit-tracker">habit tracker</a>

GAP Goals: The Three Pillars

The point of a goal is not to create a momentary-change, the point of a goal is to create and sustain a lifetime-change.The point is not “to make it” or “arrive” at an outcome, the point is to progress throughout. We don’t grow and expand the moment we reach the result, we grow and expand in the process of working toward the result we want.

Focus on Progress to Create Lifetime Change

The point of a goal is not to create a momentary-change, the point of a 
goal is to create and sustain a lifetime-change. In this article, I’m 
sharing with you a more mindful approach to goal-setting that puts growth, 
awareness, and progress at the core of it, while transforming the outcome 
into a milestone to celebrate rather than a futuristic point in time to 
arrive to.

Brian Adams
@brianadam

Better AI Systems

Our approach to building transparent and explainable AI systems

The Traditional Approach: Outcome-Based Goals

A More Mindful Approach to Goal-Setting

Brian Adams@brianadam

Better AI Systems

Our approach to building transparent and explainable AI systems

The Traditional Approach: Outcome-Based Goals

A More Mindful Approach to Goal-Setting

Brian Adams
@brianadam