Barry O'Sullivan

speaker

108 appearances 1 recordings 1 series first heard Aug 2026 last heard 14 Aug

Barry O'Sullivan’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 1 in all, peaking in Aug 2026 with 1.

Appearances

newest first · ▶ plays the moment

Taking Stock · Circular economy, 'dead internet' theory & blockchain revolution · 14 Aug 2026 podcast

You know, you're selling um you know, you were you're selling um

uh solar eclipse glasses over the last few days and all of a sudden, you know, a huge number of agents start trying to buy glasses from you.

They might be legitimate, but they might also just be trying to prevent people from accessing l legitimately your service so that they can do business.

So it is in everybody's best interests that we know that there is um that there are actual users and that there's actual um bona fide businesses behind um this traffic.

So um so all of those that you mentioned, these are real these are great opportunities.

I think another great opportunity is understanding the provenance of content and the provenance of information.

And what I mean by that is um

i if if you see content online, you should be able to verify that it is real content.

That for example it it is it is news content and it has been produced by a by a reputable journalist.

If the journalist is not reputable, at least you know that and you can you can tell that or if you know that the this journalist has a particular b a particular political bias, you can you know that that the provenance is there.

You can explore who is this person, what's this back what's the person's background and

in terms of their views on things.

Equally

When you see an when you see a website, um it's important to know that this website was, you know, how was this created?

Is this is this real information?

Um and so th this this provenance of information idea is really, really important because um what we know is um if you train AI systems on

AI generated content.

Um there's a a phenomenon called model collapse.

So very small amounts of AI generated text, for example, if included in the training sets of um large for for large language models, the performance of the large language model deteriorates very, very badly.

It's kind of like a a digital CJD, if you want.

Showing 81–100 of 108 · page 5 of 6