Watching Chasing Utopia
16 May 2026
Our new Chief AI Officer, Kalbir, recently asked us how AI might be used to drive human flourishing. It's a really good challenge; one prompted by a viewing of Mo Gawdat's new documentary Chasing Utopia. It happened to be showing in my local cinema, so I popped along to see it.
I was a little resistant at first, as the usual gang of ageing AI dudes in leather jackets lined up to warn us of the impending apocalypse. They have good intentions but they also have a habit of essentialising the tech. AI will do this, AI will do that. So often they situate the risk inside the technology, rather than with the people who control it and the systems within which it operates. I was listening out for the word 'capitalism' and didn't hear it once.
But that's my one major gripe - as the doc proceeded Mo covered off a lot of what I was hoping to hear about, including the issue of who controls the tech. I was particularly taken by Sherry Turkle's insights, about what prolonged exposure to maximally compliant human simulacra might be doing to our relationships with real people.
If it pulls a few punches it's probably because this is a doc intended to influence an audience of world leaders. And hopefully it will.
But back to that challenge that Kalbir set:
What if the best way to use AI to encourage human flourishing is to convert AI efficiency gains into free time that workers can use as they want? Perhaps this is how we get to a 4 or 3 day work week? By stealth.
What I like about this approach is that we let people decide how they want to flourish. We don't try to insert AI into every facet of their lives. Sometimes we just need to get out of the way.
But even as I write this I can feel my protestant work ethic kicking in, judging me. It's taboo, heretical, to talk like this. People who think like this are lazy freeloaders are't they? Or just naive. But then I'm sure the same was said about people advocating for weekends.
But what if we started to normalise using AI like this? its hard to imagine that happening in the private sector (capitalism again) but perhaps the real public sector AI innovation could be our enlightened attitude to how we reap the benefits of this new technology.
I think a few things would happen if we tried.
First, I think we'd discover that some people have already been doing this to some extent, but for obvious reasons they haven't been talking about it much. Have you noticed how quiet Slack can be on Fridays for example?
Secondly, I think we'd discover that many people would still choose to use that time in ways that directly or indirectly benefit the organisation. Particularly if the work is meaningful.
I also think people's ability to take advantage of AI to free up their time will be very unevenly distributed across the workforce, even within the same team. So there is going to be inequity, and resentment. We'd need to address that.
Finally, I think we'd discover that it's actually really hard to measure efficiency gains, and even harder to attribute them to specific people. That's not how work works. It's the collective endeavour of a whole organisation.
Obviously, some bosses would hate this plan. They want to convert those efficiency gains into profits. Or, in the case of the public sector, impact.
So how about this. A new deal:
- Efficiency gains made by an organisation fixing broken processes can be used by the organisation
- Efficiency gains made by staff working faster and more effectively can by used by staff
Again, I could be acused of naïveté here. The distinction between fixing a broken process and working more efficiently is fuzzy, as is the distinction between an organisation and its staff. It'll be contested ground, but at least we'd be having the debate.
So, how does an organisation fix it's broken processes? Well, not necessarily with AI.
My experience of the civil service tells me that many of the things that slow us down and make us less effective are the same things that will slow AI down and make it less effective.
A lack of focus. Competing priorities and teams. Badly defined goals. Badly defined accountability. Misaligned incentives. Poor interoperability between data and systems. Capture by the private sector. Disruptive politics.
AI alone doesn’t solve these issues but it might be a catalyst for finally addressing some of them. It would be ironic (and somewhat annoying) if it took the ‘needs’ of some emerging tech to do what decades of people complaining about it could not.