Jun 4, 2026 · 55 min · 11 segments
My guest, Fabio Tollon, a postdoctoral researcher on the BRAID programme at the University of Edinburgh, argues that answering that question is more difficult than it first appears. Traditional…
Fabio TollonGuestReid BlackmanHostLet's say I'm driving my car, I switch lanes, and this causes someone behind me to unreasonably, or for some reason, they panic, they swerve.
I did something that led to, let's say, someone dying, but I wasn't aware that I was engaged in the kind of activity that was soon going to lead to someone dying.
In that case, since I'm not aware and couldn't have reasonably been thought to be aware, expected to be aware that my changing lanes would lead to that outcome, it would be odd, it'd be weird to hold me morally responsible.
Similarly, if I was walking around and someone bumped into me and that caused me to bump someone else and then they fall, they break their leg, I wasn't in control of my actions when I got bumped into that other person.
So again, it would seem odd to hold me accountable or responsible or blameworthy for such a thing.
It looks like in order for us to be morally responsible, for us to be legitimately blamed for wrongdoing, it has to be the case that we're both aware of what we're doing and we're in control of what we're doing.
While that seems like a reasonable view, when we talk about releasing AI agents or developing AI agents that go off and do their own thing, if they go off and do crazy stuff that we didn't predict and we're not in control of because we can't monitor them at every second, it sure looks like we're morally responsible.
If I develop an AI agent and I release it onto the internet and it causes all this harm, it sure looks like I'm responsible for this thing, even though... after I've released the agent into the wild, I could no longer control the agent.
How do we square how we standardly think about moral responsibility in that it requires us to be aware of what we're doing and in control of what we're doing? And...
At the same time, also think that we're morally responsible for these agents that we can neither predict or control.
Read the full transcript.
Create an account to read the whole episode, search across every transcript, and follow the shows you care about.
Let's say I'm driving my car, I switch lanes, and this causes someone behind me to unreasonably, or for some reason, they panic, they swerve.
I did something that led to, let's say, someone dying, but I wasn't aware that I was engaged in the kind of activity that was soon going to lead to someone dying.
In that case, since I'm not aware and couldn't have reasonably been thought to be aware, expected to be aware that my changing lanes would lead to that outcome, it would be odd, it'd be weird to hold me morally responsible.
Similarly, if I was walking around and someone bumped into me and that caused me to bump someone else and then they fall, they break their leg, I wasn't in control of my actions when I got bumped into that other person.
So again, it would seem odd to hold me accountable or responsible or blameworthy for such a thing.
It looks like in order for us to be morally responsible, for us to be legitimately blamed for wrongdoing, it has to be the case that we're both aware of what we're doing and we're in control of what we're doing.
While that seems like a reasonable view, when we talk about releasing AI agents or developing AI agents that go off and do their own thing, if they go off and do crazy stuff that we didn't predict and we're not in control of because we can't monitor them at every second, it sure looks like we're morally responsible.
If I develop an AI agent and I release it onto the internet and it causes all this harm, it sure looks like I'm responsible for this thing, even though... after I've released the agent into the wild, I could no longer control the agent.
How do we square how we standardly think about moral responsibility in that it requires us to be aware of what we're doing and in control of what we're doing? And...
At the same time, also think that we're morally responsible for these agents that we can neither predict or control.