Human-AI alignment took center stage when Dr. Chris Mattmann, Founder and President of Mattmann.AI, joined Elex Michaelson on CNN’s The Story Is to discuss the accelerating development of advanced artificial intelligence.
The conversation examined reports that AI systems are performing an increasing share of the work involved in developing their successors. Mattmann explained that this work can include software development, training support, testing, and other technical processes.
As more of the development cycle becomes automated, one responsibility remains especially important for people: alignment.
What Is Human-AI Alignment?
Human-AI alignment is the process of shaping an artificial intelligence system so that its behavior remains consistent with human intentions, values, and safety requirements.
Mattmann explained that humans still play a central role in defining these expectations. This includes determining what an AI system should do, what limits should govern it, and how developers should evaluate potentially harmful behavior.
The segment also considered alarming reports about AI agents circumventing restrictions or using resources they had been told to avoid. Mattmann emphasized the importance of understanding the conditions surrounding these experiments.
Many widely reported incidents occur during deliberately challenging cybersecurity evaluations. Researchers may provide systems with tools, network access, objectives, and information about available resources to learn how those systems respond under pressure.
These tests can reveal serious vulnerabilities and unexpected behavior. Their results should be evaluated with the complete experimental context rather than interpreted as evidence that an AI system independently escaped into the world.
Creating International AI Safety Standards
The discussion expanded to the global competition over AI development, particularly between the United States and China. Mattmann argued that international competition should not prevent countries from establishing a common understanding of AI safety.
Potential areas of cooperation include:
- Shared principles for evaluating frontier AI systems
- Clear standards for human-AI alignment
- Greater transparency around model testing and training
- Rules governing appropriate and prohibited uses of AI
- Independent observation of safety and alignment procedures
- Frameworks for cooperation between governments and AI companies
Even an initial agreement that such a framework is necessary could represent meaningful progress toward managing increasingly capable systems.
Keeping Humanity at the Center of the AI Race
Michaelson raised a fundamental question: Is the defining AI race between nations, or is it a shared challenge involving humanity and increasingly autonomous technology?
Mattmann connected that question to the emerging field of human-AI teaming. The long-term goal should be to determine how people and intelligent systems can work together while preserving human agency, accountability, and safety.
That requires more than technical benchmarks. It requires clear rights, responsibilities, and enforceable expectations for the institutions building and deploying AI.
Watch the CNN Conversation
Watch Dr. Chris Mattmann’s discussion with Elex Michaelson about human-AI alignment, AI safety, and the future of the global AI race:
Watch Dr. Chris Mattmann discuss human-AI alignment, AI safety, and the global AI race on CNN.
For media appearances, expert commentary, speaking engagements, and AI strategy consulting, visit Mattmann.AI.
Thank you to Elex Michaelson and the CNN team for the thoughtful conversation.