The Role of Explainable Interfaces and Building Trust in Government

As artificial intelligence (AI) becomes increasingly integrated into government services, building public trust and ensuring transparency in AI-driven decision-making processes have emerged as critical challenges. To address these concerns, explainable interfaces are playing a crucial role in enabling transparent AI and fostering trust in government AI systems.
In this talk, Noble Ackerson, a Fractional AI Risk Manager and Chief Product Strategist at Byte an Atom Research, will explore the importance of trust and transparency in adopting AI within government services and how explainable interfaces can help achieve these goals. By adhering to Paul Grice's maxims of effective communication, these interfaces ensure clear, relevant, and cooperative interactions between users and AI, making AI more accessible and user-friendly compared to traditional chatbots.
Noble will present real-world examples of government agencies using explainable AI, enabled by intuitive interfaces, to provide users with interpretable insights from predictive and prescriptive data on dashboards. These case studies will demonstrate how explainable interfaces can foster public trust and engagement by demystifying AI decision-making processes and promoting accountability.
Finally, Noble will discuss the future of AI in government, and the critical role explainable interfaces will play in ensuring responsible and transparent AI deployment. Attendees will learn about the challenges and opportunities associated with implementing explainable AI in government services and gain insights into how they can leverage these interfaces to build trust in their own AI initiatives.
Attend this GovAI Summit talk to explore the cutting-edge of explainable AI and discover how explainable interfaces are transforming the relationship between government, AI, and the public.

Building Explainable Interfaces for your Apps

In this talk, Noble Ackerson, a Fractional AI Risk Manager and Chief Product Strategist at Byte an Atom Research, will explore how to make user interfaces smarter with Large Language Models (LLMs)
As AI becomes increasingly integrated into our applications, it's crucial to rethink how we design user interfaces. Traditional chatbot interfaces place an undue cognitive burden on users, expecting them to become "prompt engineers" to effectively communicate with AI, leading to potentially harmful responses, hallucinations, and unreliable outputs.
What if we could make websites smarter with AI? Noble will walk through how he creates smarter, more intuitive interfaces for AI-powered apps.
This code lab-style talk applies the principles of cooperative communication and leveraging the power of Large Language Models to enable context-sensitive, real-time assistance that adheres to Paul Grice's maxims of effective communication.
We'll explore how his approach abstracts prompting from users, fosters clear and relevant interactions between humans and AI, and ultimately delivers a more seamless and satisfying user experience. Attend this Voice&AI talk to discover the future of AI-driven interfaces and learn how to build smarter apps that truly harness the potential of AI.

AI's Complexity Trap: Unpacking the Operational Complexity Paradox

What if our efforts to improve the reliability of generative AI solutions are making these systems more cumbersome to build and hard to use? I call this the "AI Complexity Trap." As we add more layers of safety and accuracy, we're inadvertently creating complex systems that are draining resources and complicating things.

In this talk, I'll expose the hidden costs of over-engineering AI and share solutions for balancing excellence with simplicity. You'll discover strategies to harmonize technological advancement with pragmatic management. The goal? Harnessing AI's potential without falling into the complexity trap.

GenAI: An Unreliable Information Store...and What to Do About It

Synopsis: Embark on an enlightening journey with Noble as he tackles the challenges of integrating Large Language Models (LLMs) into enterprise environments. Understand the inherent unreliability of these models and explore innovative solutions, ranging from vector databases to prompt chaining, that aim to enhance the trustworthiness of LLMs in crucial applications. About Noble Ackerson: An authority on data ethics and emergent technology, Noble is at the forefront of applied AI integration at Ventera Corporation, Virginia. His illustrious career boasts accolades as an AI Product Manager, Google Design Sprint Master, and former Google Developers Expert. With a legacy of crafting award-winning mobile and advanced analytics solutions, Noble's passion extends to mentoring underrepresented communities. [Read more](https://medium.com/@nobleackerson)

Trustworthy AI for Enterprise: Enhancing LLMs

GPT Is an Unreliable Information Store
Current large language models (LLMs) exhibit limitations in deductive reasoning and cognitive architecture, posing challenges to their reliability in enterprise applications. This talk presents novel techniques to address these limitations, evaluate, and improve LLM performance.

We will highlight the issue of epistemological blindness in LLMs and propose a solution using embeddings model endpoint and Vector Index DB for enhanced factual accuracy. Additionally, we will discuss innovative approaches to automate feature engineering, including Automatic Feature Extraction with LLMs and Langchain integration in an MLOps pipeline so that we can evaluate the accuracy of these tools.

This presentation aims to offer valuable insights for practitioners and researchers, contributing to developing responsible and trustworthy AI in enterprise contexts.

Tags: Explainability, ML Model Pipelines, Chroma, Pinecone, OpenAI, ada, Embeddings, Cosine Similarity, Semantic Search

Fight AI bias with… bias

How can we ensure that AI systems treat everyone fairly?

Human biases influence the outputs of an AI model. AI amplifies bias, and socio-technical harms impact fairness, adoption, safety, and well-being.
Harms such as improperly implemented AI-driven predictive policing disproportionately affect legally protected classes of individuals and groups in the United States.

In this talk, Noble walks through tools/libraries to mitigate bias within your ML pipeline and explainability solutions with SHAP for explaining the predictions of machine learning models to avoid disproportionate product failure due to fairness and bias issues.

It’s so fitting that the 2022 theme for International Women’s Day was #BreakTheBias, so join Noble as he returns to Strangeloop to expand on the topic of bias, deconstruct techniques to de-bias datasets by example for building intelligent systems that are fair and equitable while increasing trust and adoption.

Tags: Explainability, XAI, Fairness Metrics, ML Evaluation, Model Monitoring, Interpretable Models, Machine Learning, ML/AI

Defining Data Trust in XR - IAPP Global Privacy Summit 2022

Defining Data Trust in Extended Reality @ Noble joins Rob Sherman, Deputy CPO at Meta and Nicol Turner-Lee, Fellow, Center for Technology Innovation, Brookings Institution
To discuss privacy in XR at the IAPP Global Privacy Summit 2022

Technical constraints and privacy considerations for #UNID2020 mission

During the 63rd Session of the UN Commission on the Status of Women, I was invited to join a panel of thought leaders and international practitioners on the impact and opportunities the UN "ID for all" presents and, from a technical perspective, address the intersection of technical feasibility, standards, and privacy regulations such as the EU's GDPR.

https://www.ndi.org/international-womens-day

ARCore Parkour: Incorporating marker-based AR into your app

Data gathered through Augmented Reality, and vision camera systems help computers bridge information from the real and virtual worlds so that one understands the other.

Augmented reality (AR) and mixed reality (MR) were once technologies considered reserved for games and media entertainment. However, these emerging technologies are growing rapidly as many industries realize that the potential of these technologies can be game changers. These technologies represent potential new leaps in how business is run, data is collected, and even new value-add propositions.

In this talk, Noble Ackerson, a 7-year Augmented Reality Developer and Product Expert, shares development best practices and industry insights, with examples such as incorporating marker-based AR into your app and creating XR experiences that cannot happen anywhere else.

Product Innovation in a Privacy by Design World

Personal data privacy is a hot-button item today. It can often seem overly burdensome for the software developer looking to change the world. Established businesses may look at regulations and current data privacy trends as negatively impacting product innovation.

This is the wrong lens through which we should look at this.

In this talk, Noble walks through his framework to respect user privacy AND innovate around the growth in data, and the growth in regulations like the GDPR.
He shares insights on human-centered design opportunities that can help the independent developer and an established corporation.
Finally, he presents practical examples to minimize the data you collect and best practices for asking for data in software only when it provides value for the user.

AWE: Shadows of the emerging worlds

Today we see advancement in practical applications of emerging technologies in our homes and at work. As these solutions grow towards ubiquity, so do the possible harms and obstacles to the world we all want to live in.

In this talk, Noble Ackerson walks through 3 data trust and ethics issues in XR specifically:

1. Dark patterns in XR - impact, example, opportunity;
2. AI fairness & bias - impact, example, opportunity;
3. Data privacy/trust and security - impact, example, opportunity.

He presents practical examples of principles to avoid wading into dark patterns within the shadows of the emerging world and how you optimize for trust through his explainability, context, control, and choice framework for good data stewardship.