Welcome back to The Harvard AI Diaries, day four. Today was packed with big topics and even bigger questions—especially the kind I hear all the time from clients in government and the public sector.
Copyright. Intellectual property. Misinformation. Disinformation. The future of work. And agentic AI.
Let’s start with copyright and IP. Sharad Jial, who’s leading the programme, put us through our paces this morning. The session was electric. So many hands up. So many questions. Including mine, several times. He eventually had to say, guys, we need to move on.
But the big three questions he posed were ones that come up all the time in my training sessions:
- Does the large language model training process infringe copyright?
- Can AI-generated outputs be copyrighted?
- Do those outputs infringe the copyright of someone else’s work?
Huge, thorny questions. No easy answers. And I’m not going to try to solve them here in six minutes. But what made it real was diving into live case studies—including the OpenAI vs New York Times case. We actually had to take sides and argue our position using ChatGPT. That kind of engagement changes how you think.
We then shifted into misinformation and disinformation—topics I know are keeping public sector leaders up at night.
Here’s a simple distinction:
- Misinformation is incorrect info shared without intent to mislead.
- Disinformation is intentional, often politically or commercially motivated.
We examined real-world examples, especially during elections, and asked: Can we trust large language models like ChatGPT, Gemini, or Meta’s Llama?
There are mitigating strategies. I actually came away a bit more optimistic because we saw examples of how trust can be built through transparency, labelling, and governance.
One of the most memorable contributions came from Teddy Savoris, who introduced the “liar’s dividend”—a concept I hadn’t heard framed quite like this before. The idea is that you don’t need much disinformation to do damage. A little can go a long way. Researcher Hany Farid has shown that once the well is poisoned, people start to doubt everything. And that’s a problem, because when AI-generated content floods the space, even real content becomes suspect, especially if it challenges someone’s existing beliefs.
We then debated the roles and responsibilities of social media companies, AI developers, civil society, and government. And it became clear that the extent to which generative AI worsens these problems depends entirely on how we, collectively, respond.
Then came a session on agentic AI, a topic I’ve mentioned on this podcast before. If it’s new to you, think of AI agents as tools that don’t just generate text or images; they act. They complete tasks. They make decisions. They run autonomously within guardrails. And they’re already changing how we think about productivity and workflows.
And, perfect timing, today, as we were learning about AI agents here at Harvard, OpenAI launched their own GPT Agents. A real-world example unfolding in real time. It was one of those moments where theory and practice collide.
We also explored the future of work, how generative AI is reshaping not just jobs, but entire career paths. There are big questions around inequality, too. Who gains from this technology? Who gets left behind? What kinds of skills will matter in the next five to ten years? These are questions I hear from public servants and policymakers across Europe, and they were very much on the table today.
So to wrap up day four: the conversation is shifting from hype to hard questions. What’s legal? What’s ethical? What’s safe? And how do we build systems that protect people, preserve rights, and promote trust?
That’s it for today. I’ll be back tomorrow with more reflections from Harvard and a look at how we’re being asked to lead in the age of AI.
About the Podcast Host
Joanne Sweeney is a specialist in public-sector marketing and digital communications. She’s the author of Public Sector Marketing Pro and the creator of the Professional Diploma in AI for Public Sector Marketing. With years of experience in training, consulting, and advocating for digital transformation, Joanne provides insight and guidance for government agencies navigating the digital landscape. She is also the host of the award-winning Public Sector Marketing Show podcast.
