Toyota’s Manufacturing Agentic Journey

Team Toyota has strategically leveraged Generative AI to drive innovation and operational excellence across key domains, including manufacturing, battery plants, paint design, and enterprise knowledge management through solutions like
Knowledge Bot (Kura) and Global Chatbots. By integrating Large Language Models (LLMs), Toyota has developed a robust methodology to evaluate, score, and quantify AI outputs against Toyota’s quality benchmarks, ensuring high standards and reliability in production environments.

This effort highlights critical lessons in collaborative requirement discovery and outcome-defined capability selection, enabling efficient resource utilization and fostering strong enterprise leadership engagement. These practices ensure AI initiatives align with business objectives, accelerating adoption and delivering measurable value across the organization.

A cornerstone of this success is the application of the agentic framework for deploying Generative AI in production, enabling autonomous AI agents to execute complex tasks and optimize processes seamlessly. This approach has transformed workflows, enhanced decision-making, and driven significant operational efficiencies.

Toyota’s strategic deployment of Generative AI demonstrates a commitment to innovation, showcasing how advanced technologies can be effectively utilized to maintain quality, scale operations, and strengthen leadership in an evolving digital landscape.

Generative AI Developers, Executives & Researchers.

Algorithmic Bias and Discrimination in AI: Developing Regulatory Frameworks to Mitigate Harm in Africa

This research investigates the manifestations of algorithmic bias and discrimination in AI systems deployed in African contexts.
It explores how biases embedded in data, algorithms, and the socio-cultural contexts of AI development can lead to discriminatory outcomes.

• Identify and categorize biases in AI systems deployed in Africa.
• Analyze the impact of these biases on individuals and communities.
• Explore methods for identifying and mitigating algorithmic bias.

Addressing Algorithmic bias, Data bias, and Societal bias.

• AI systems deployed in Africa are susceptible to racial and gender biases.
• Biases can perpetuate existing inequalities and harm communities.
• Mitigating bias requires fair data collection, unbiased algorithms, and continuous monitoring.
• A collaborative effort among researchers, developers, policymakers, and communities is necessary.

Ethical and Responsible Autonomous AI for Human-AI Interaction

Professor Keeley Crockett

Manchester Metropolitan University, UK Leads the Machine Intelligence theme in the Centre for Advanced Computational Science. She has over 27 years of experience in Ethical and Responsible AI, computational intelligence, psychological profiling, fuzzy systems, and dialogue systems. She is a member of the IEEE Computational Intelligence Society ADCOM (2023-25) and chairs the IEEE Technical Committee SHIELD (2023-24).

Professor Ricardo Baeza-Yates

Northeastern University, USA Research Professor at the Institute for Experiential AI. Former VP of Research at Yahoo Labs (2006-2016). Co-author of the best-seller Modern Information Retrieval. Fellow of ACM (2009) and IEEE (2011). Expert in AI bias, data science, and algorithms, actively involved in global AI ethics initiatives.

Professor Yew-Soon Ong

Nanyang Technological University, Singapore President Chair Professor in Computer Science at NTU and Chief AI Scientist of the Agency for Science, Technology, and Research Singapore. Founding Editor-in-Chief of IEEE Transactions on Emerging Topics in Computational Intelligence. His research focuses on AI and computational intelligence. Recipient of multiple IEEE outstanding paper awards and listed among the World’s Most Influential Scientific Minds.

Professor Jim Torresen

University of Oslo, Norway Leads the Robotics and Intelligent Systems research group. Research areas include AI ethics, machine learning, and robotics. Has published over 300 peer-reviewed papers, delivered 48 invited talks/keynotes, and organized international AI conferences. Member of the Norwegian Academy of Technological Sciences (NTVA) and IEEE RAS Technical Committee on Robot Ethics.

• Design phase: Developing fair and responsible AI systems.
• User decision-making: When and where AI systems should be deployed.
• Operational phase: Ethical reasoning in autonomous AI systems.

Perspectives on Current Global AI Governance Trends and the Way Ahead for 2025

This panel highlights global developments in AI governance over the past year in the United States, Europe, and the Asia-Pacific.
The discussion will explore relevant issues and key questions for AI governance in the coming year.

Since 2023, major AI advancements have prompted global governance responses:

• The U.S. issued new executive orders and policy frameworks to ensure AI security and trustworthiness.
• The European Union finalized its AI Act and is drafting a code of practice for general-purpose AI.
• China launched its Global AI Governance Initiative, shaping international AI regulatory strategies.

This panel will evaluate the progress of these initiatives, highlighting key synergies, gaps, and areas of tension.
The discussion will focus on how the U.S., EU, and China manage AI risks and benefits, and what these governance trends signal for AI in 2025.

This panel is aimed at researchers and practitioners tracking AI governance trends, as well as technology developers interested in understanding
governance impacts across different regions. The discussion will offer a comprehensive overview of AI policy developments over the past year.

Great Infrastructure Debate: Choosing the Right AI Backbone Amidst Chaos

With the rapid emergence of new LLM models and diverse cloud infrastructure options beyond traditional CPUs and GPUs,
selecting the right AI infrastructure is a crucial and contentious decision. This panel will explore strategic AI deployment
across industries including healthcare, transportation, manufacturing, and business intelligence.

1. Infrastructure Battles: Evaluating cloud, on-premises, and hybrid models with insights from AWS, Google, Microsoft, CoreWeave, and Vultr.
2. Data Sovereignty: Addressing storage, security, and compliance across platforms.
3. Accelerator Wars: Comparing GPUs, TPUs, and emerging solutions like Groq, Furiosa AI, and Cerebras.
4. Scalability vs. Sustainability: Balancing TCO, resource optimization, and sustainable AI growth.
5. Model Mayhem: Choosing and fine-tuning AI models, emphasizing multimodal LLMs and integration with DevOps/SecOps.
6. Build or Buy: The pros and cons of in-house AI development versus AI as a service.
7. Monetizing AI: Exploring ROI strategies for enterprises and hyperscale AI operations.
8. True Costs Unveiled: Analyzing total AI deployment costs, including infrastructure, personnel, and market readiness.

This panel will engage in a dynamic discussion about AI infrastructure trends and challenges. Attendees will gain insights
into aligning AI initiatives with the most suitable infrastructure strategies, optimizing for both current capabilities and
future demands.

Building Trust for Human-AI Partnerships in Security

Cybersecurity offers a rich case study in human-AI collaboration with lessons that apply across domains. Through real-world examples of AI integration in threat intelligence, detection, and response, this panel explores how teams build effective partnerships with AI systems. We examine practical approaches to reducing cognitive load, improving decision-making, and creating sustainable workflows that benefit technical specialists and broader stakeholders.

Security teams face increasingly complex challenges that demand effective collaboration between humans and AI systems. This panel examines practical approaches to building successful human-AI partnerships through real-world examples from threat intelligence, security operations, and incident response.

Our diverse panel brings together academic research, operational experience, and product development insights to address key challenges:

• Designing interfaces that adapt to different expertise levels without compromising effectiveness

• Implementing explainable AI that builds confidence in automated decisions

• Creating workflows that reduce analyst fatigue while maintaining human judgment

• Developing clear metrics for evaluating AI system trustworthiness

• Setting realistic expectations for AI capabilities across different security domains

• Building sustainable practices that prevent burnout and enhance team resilience

Drawing from direct experience integrating AI across security functions, panelists will share specific strategies for:

• Evaluating when and how to trust AI-generated insights

• Establishing effective feedback loops between humans and AI systems

• Measuring impact on team performance and analyst wellbeing

• Scaling AI benefits across different security roles and expertise levels

Attendees will learn specific methods for evaluating AI security tools, clear criteria for assessing AI system reliability, and proven techniques for integrating AI assistance without overwhelming analysts.