Advances in Generative AI

University of Colorado Boulder via Coursera

Go to class Write review

Overview

Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off

One annual plan covers every course and certificate on Coursera. 40% off for a limited time.

Get Full Access

In this course, you’ll learn how generative AI systems evolve from tools into more autonomous, goal-driven systems—and what that means for how they are built, evaluated, and used in the real world. You’ll explore how foundational models, feedback loops, tools, and memory combine to create agent-like behavior, and how modern AI systems are designed as coordinated “teams” rather than single models. Along the way, you’ll examine how AI is being applied in areas like scientific discovery and complex workflows, while also learning how to evaluate performance, manage risk, and design systems responsibly. By the end of the course, you’ll be able to think like an orchestrator—someone who can guide, oversee, and safely deploy increasingly capable AI systems in your field.

Syllabus

Foundations, Evaluation, and the Rise of Autonomous Systems

In this module, you’ll learn how modern generative AI systems are built and how their capabilities are measured. You’ll explore foundational ideas like transformers, scaling, fine-tuning, and reinforcement learning, and see how these shape what models can and cannot do. You’ll also examine how AI is evaluated—through benchmarks, human judgment, and long-horizon tasks—and why strong scores don’t always translate to real-world reliability. Finally, you’ll explore how feedback loops enable systems to improve over time, and begin thinking about when (and if) these systems should be trusted to act more autonomously.

What Makes an Agent an Agent: From Tasks to Objectives

In this module, you’ll learn what makes an AI system an “agent” rather than just a tool. You’ll explore how agents operate over time by combining reasoning, memory, tools, planning, and verification into a continuous loop. You’ll also learn how agents “sense” and respond to changing information, and why that matters for real-world applications. A key focus will be thinking of agents as coordinated teams—with roles like planner, executor, and evaluator—and understanding where human oversight must remain in place. By the end, you’ll have a clear mental model for how agents differ from prompts and workflows.

Agent Architectures, Multi-Model Workflows, and Orchestration

In this module, you’ll learn how agent systems are designed in practice. You’ll explore how different roles—like planning, reasoning, tool use, and evaluation—are structured into working systems, and why many real-world solutions rely on multiple specialized models instead of just one. You’ll examine how these systems are orchestrated, how tasks are routed between components, and how coordination affects performance, cost, and reliability. Through examples and case studies, you’ll shift from thinking about prompts to thinking about systems—and learn why orchestration and verification are the key skills for advanced AI use.

The Limits of Agents Today: Verification, Security, and Governance

In this module, you’ll learn where today’s agent systems fall short—and why human oversight is still essential. You’ll explore common limitations like weak long-term planning, unreliable memory, and alignment challenges, and understand why autonomy does not equal understanding. You’ll also learn how to design safer systems by using verification, permission controls, and “guardrails” that limit what agents can do. Beyond the technical side, you’ll examine broader risks like bias, misuse, and security vulnerabilities, and learn how governance and responsible design play a critical role as AI systems become more capable.

Scientific Discovery, Simulation, and Emerging Research Directions

In this module, you’ll learn how generative AI is being used beyond productivity—to accelerate scientific discovery and innovation. You’ll explore real-world examples in areas like biology and materials science, and see how AI can support hypothesis generation, simulation, and experimentation. You’ll also be introduced to emerging ideas like world models, which combine memory, simulation, and planning to enable more advanced reasoning. Rather than focusing on predictions about AGI, this module will help you understand the building blocks of more general capabilities and how to interpret ongoing research trends.

Working Alongside AI in an AI-Augmented Society

In this module, you’ll learn how to position yourself in a world shaped by increasingly capable AI systems. You’ll explore where human skills—like judgment, oversight, coordination, and ethical decision-making—remain essential, even as automation increases. You’ll revisit the idea that AI capability often advances faster than adoption, and learn how that gap creates opportunities for those who can safely deploy and manage these systems. By the end, you’ll develop a clearer sense of how to work alongside AI strategically—focusing not on competing with it, but on using it to enhance your value.