Today's students will have a 40yr career with powerful AI agents working for them, not simply chatting, but acting on complex issues. Whichever field of work they enter, everything from film production to finance will be employing AI. Managing that AI will be analogous to managing people. By that logic, they will need these management skills:
Ensure the AI is working in their best interests
called 'alignment' in AI
Hire and manage AI effectively
called 'evaluations' in AI
The majority of our students will enter some form of business and many will manage their own team or even company. AI Agents already support customers, write code and they can even hire people for the tasks they can't do. Thanks for that last one VISA, a little ominous even for me!
Our young people will be heading up teams of these AI Agents and, similar to managing people, will face two deceptively simple tasks:
Specify goals correctly
Get AI to follow those goals

What Could Go Wrong?
Goal setting sounds trivial. Yet, they are a moving target, they shift as we get better sight of them, aka 'course correcting'. Furthermore, setting simple goals leads to 'gaming the system'. Think of the NHS chasing simple reporting targets, not complex patient outcomes.
Ensuring employees and agents follow your goals is even more complex. Agents will rarely have only one agenda, usually they have multiple masters; you, their vendor, a regulator, and even an interloper. As we already know, the relationship between humans acting as agents and their 'principals' has a long history of breaking down in unexpected ways. An entire field of law has developed to govern it, agency law, a subset of contract law. There are already proposals on how it adapts to AI.
As with managing people, today's students can expect to meet AI employees who are 'saints', some are 'sycophants' and others 'schemers'. Yes, it sounds unlikely, but it already happens on social media and the UK Govt's AI Safety Institute (AISI) already test models for deception, much as regulators insist upon banks who 'know your customer' to prevent fraud.
Given this background, researcher Ajeya Cotra underlines the magnitude of the eye watering problems with this thought experiment:
Imagine you are an orphaned eight-year-old whose parents left you a $1 trillion company, and no trusted adult to serve as your guide to the world.
You have to hire a smart adult to run that company, guide your life the way that a parent would, and administer your vast wealth. You have to hire that adult based on a work trial or interview you come up with.
You don’t get to see any resumes or do reference checks. And because you’re so rich, tonnes of people apply for the job — for all sorts of reasons.

Master of Agent Administration *
Given the impact AI will have on business, it’s not too far-fetched to imagine the students of tomorrow completing their studies with a Masters in Agent Administration instead of a Masters in Business Administration.
We will be dropping our students into management positions from day one. They need to learn more than how to do business. They must oversee the AI agents who are running the business for them, even if working alone.
The closest we currently have is the course I recently completed. It's aimed at software developers, researchers and experienced people in AI or governance. It's 12 weeks, but taught by real researchers (not recorded, no ai agents) and within a small tutor group.
BlueDot Impact AI Safety Fundamentals: Alignment
The sister course is for those seeking careers in law and policy:
BlueDot Impact AI Safety Fundamentals: Governance

How to Get Started with AI Alignment
Not Ready for a 12 week course?
The most digestible introduction to these issues is Robert Miles' classic video from the pre ChatGPT era, three years ago.
Be warned that like other transformative technologies, AI can be used for good or for ill. The video does not shy away from discussing the threats, these are the call to arms for AI Safety.
What Next?
Looking for expert knowledge in AI Agents? At Agentico, we are specialists in this technology, to find out more, book a discovery call
Oliver
Agentico Ltd.
AI with (your) Humans in Mind
* Master of Agent Administration - trademark pending.
Comments