Usually I put on a podcast to help me sleep, but last nights “Rest is Politics” kept me awake and full engaged, as they had one of Anthropics’s co-founders Jack Clark being interviewed. I was half expecting another puff pro-AI marketing piece (especially since they filed their IPO today), but Jack Clark (a fellow Brit) surprised me.

Rory and guest host Matt Clifford (founder of the UK AI Safety Council) grilled him on some very challenging questions – the usual ones about AI destroying human civilization – and Jack to his credit, said yup that is a real problem and one that genuinely concerns him. Okay then – you have my interest. Carry on.

Clark talks through, calmly and thought through, how most sciences through their journey to breakthrough will discover all the things that can be repurposed for bad things. He cites biochemistry as an example, that all the good has come out of figuring out what makes us healthy, we’ve also discovered how to make bio-weapons. He believes governments have a role to police and set the standards to which all need to follow, even though they may differ slightly in different parts of the world. I liked his airplane analogy, where all governments have agreed upon a common standard as to what makes flying commercial planes safe – without that, we would never trust taking off in a foreign country.

I am bullish on AI, but I am also a realist, and don’t get caught up in the hype. I can’t stand the AI psychosis that seems to capture some management, who are blindly running employees to a place they are not completely bought into, throwing those aside like don’t buy into their utopia viewpoint, looking to reduce headcount.

Clark addresses this, noting how AI is an accelerant and organizations need to find their level of comfort adoption and believes this is going to be a net producer of jobs. He believes this will open new areas of opportunity that we haven’t yet thought about. On this point I do agree with. Yes, some of the roles we have at the moment is frankly a failing in the current technology (how many times are we moving data from one system to another that you would think should just work for example?).

This is well worth the listen/watch. He is not given an easy time (which is one of the qualities I like from Rory) but by doing so, we get a much rawer and honest answers to the questions we’re all thinking. He doesn’t pretend to have all the answers which is what I appreciated the most. The section on Mythos is fascinating.

An non-hype filled, AI is going to solve everything, discussion from the one of the people who created Claude.