We need to build the table
A kind of vertigo sets in on the third day of SXSW London. An MIT scientist compared modern societies to pilots flying without instruments. The inventor of the World Wide Web admitted on stage he is worried about what has been built on top of his invention. By the time I returned home, one thing was crystal clear to me: we are living inside the AI-experiment – with companies that increasingly know us better than we seem to know ourselves. Four takeaways for Npuls and the Dutch tertiary educational sector.
1. We are running a live experiment on an entire generation, and nobody agreed to the terms
Growing up with an always-available, endlessly patient, never-judging, conversational partner in your pocket creates the "trust paradox": the more comfortable we become with outsourcing our emotional needs to machines, the less practised we become with other human beings. AI companions don't introduce friction, and this is the most consequential design decision of our time: friction isn't a bug in human relationships. It’s how we learn to be in them.
But what hit home most, wasn't a single statistic. It was the realisation that we’re running a real-life experiment on an entire generation, with no control group, no informed consent, and no agreed endgame. It's a grand wager about what happens to human intimacy when its closest substitute never says ‘no’.
2. Whoever frames the question, governs the answer
“Most organisations are flying blind on AI,” as one speaker put it plainly. They can’t tell who their power-users are, what they’re doing with the tools, and whether any of it is working. Presented with a trade-off –either invest in AI for healthcare or in food security– the audience split two-to-one. Then the same choice was reframed: early disease detection versus AI-designed crops. The split flipped dramatically. Nothing about the actual trade-off had changed: only the wording did.
And this isn't a parlour trick, it’s governance hiding in plain sight. Every time someone says, "AI is neutral" or "this direction is inevitable", they’re making a framing choice.
The session's broader point: we are fundamentally bad at predicting where transformative technology leads. We should expect to be surprised by AI and build governance structures humble enough to admit it.
3. Voice, memory, data: who owns what pieces of you?
As part of an immersive audio performance a disturbing experience was shared: voice artist Emma Clarke was offered a contract to become the official voice of a major US-based company. Buried in the final clause of her contract was the requirement to waive all rights to her own voice, in perpetuity. The phrase was almost comic in its overreach: until the sun dies. Except, it was simply what the company wanted. Clarke found it so absurd that she agreed to read the clause aloud as final part of the performance itself, turning contract language into protest art.
Sir Tim Berners-Lee made a related argument: the www began as a good idea and became toxic along the way. Large language models represent a similar escalation: unlike social media's marketing tools, LLM’s never forget, not to mention that many of them ingested data that wasn’t lawfully theirs. The question sitting at the centre of all of this: when a piece of you –your voice, your habits, your history– becomes useful to a machine, who's is it? Yours, or the machine’s?
4. Keep it cool and bring your own seat
Sir Nick Clegg pushed back on the pessimism: eighty-five percent of the global economy is still about physical things moving around the planet –logistics, manufacturing, robotics, industrial systems. That is precisely where Europe's expertise sits. His advice: stop listening to technologists pronounce on labour markets they don't understand and recognise that real-world adoption is always slower than the hype. His words: "keep it cool."
Bo Young Lee built on this in her session: technology isn't a technical problem; it's a societal one. You don’t need to be an engineer to have a standing in this conversation. You need to be a citizen who refuses to opt out. Ask who benefits from a narrative before accepting it. Ask what is being made deliberately hard to understand, and why. If you don't have a seat at the table, it's very likely you're on the menu.
A lesson for Npuls and the education sector
If there's one thing SXSW did for me, it's strengthening my conviction in our Npuls mission: we need to make AI accessible to education on terms that serve learners, educators, and public values. The table is still being built by governments at national and European levels, whilst they're still deciding their positions. As a tertiary education sector, we need to help build this table and pull up a chair. Learners are counting on us to do so, and as Npuls and our partners, have a real opportunity to unite the public tertiary education around shared interests.