Sep 25, 2026 · 46 min · 13 segments
What on earth is happening with AI? Agents are hacking systems. Inventing their own languages. Covering their tracks and behaving like spies from a 1980s farce. The bosses of AI companies are…
Matt ArmitageGuest
Kishan SivaswamyHost
Anyway, so what do you think the first thing we've learnt over the last couple of weeks is?
After our conversation last week and based on how you started it, do you realize that we are speaking about AI, like, as though it's been there for fifteen, twenty years? But what we fail to understand is that the first commercial model for AI was actually only launched five years ago.
And from that we can learn that, um, we are starting to finally see the real questions, and that is, are we actually ready for this?
Um, and, you know, I mean, you've, you've put your, your finger on it directly.
Um, way, way back in the mists of time, a long time pre-COVID, in fact, I did a, a series of shows and public talks on artificial intelligence from the perspective of emerging intelligence and, uh, the possible possibility of those systems developing some kind of sentience.
I mean, think how priv-- primitive these research models were, uh, ten years ago.
Um, you know, even, even two or three years ago, um, experts were talking about deadlines of twenty fifty or twenty thirty-five to develop the kind of emerging general intelligence systems that we're starting to see today.
Uh, and you know, we, we talked about the, the kind of potential dangers this-
And I would argue from the position that, uh, non-sentient, you know, dumb AI could be incredibly dangerous, perhaps even more dangerous, uh, because it would just simply try to carry out whatever task you gave it, irrespective of what was happening around it.
You know, this is the, um, gray goo argument or Nick Bostrom's paperclip maximizer, um, idea that an AI without determinism, uh, could pursue a goal with, you know, such singularity that it consumes all the resources on the planet to, to meet that task.
Anyway, so what do you think the first thing we've learnt over the last couple of weeks is?
After our conversation last week and based on how you started it, do you realize that we are speaking about AI, like, as though it's been there for fifteen, twenty years? But what we fail to understand is that the first commercial model for AI was actually only launched five years ago.
And from that we can learn that, um, we are starting to finally see the real questions, and that is, are we actually ready for this?
Um, and, you know, I mean, you've, you've put your, your finger on it directly.
Um, way, way back in the mists of time, a long time pre-COVID, in fact, I did a, a series of shows and public talks on artificial intelligence from the perspective of emerging intelligence and, uh, the possible possibility of those systems developing some kind of sentience.
I mean, think how priv-- primitive these research models were, uh, ten years ago.
Um, you know, even, even two or three years ago, um, experts were talking about deadlines of twenty fifty or twenty thirty-five to develop the kind of emerging general intelligence systems that we're starting to see today.
Uh, and you know, we, we talked about the, the kind of potential dangers this-
And I would argue from the position that, uh, non-sentient, you know, dumb AI could be incredibly dangerous, perhaps even more dangerous, uh, because it would just simply try to carry out whatever task you gave it, irrespective of what was happening around it.
You know, this is the, um, gray goo argument or Nick Bostrom's paperclip maximizer, um, idea that an AI without determinism, uh, could pursue a goal with, you know, such singularity that it consumes all the resources on the planet to, to meet that task.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.