IT Essentials: Stop making sense

Suddenly powering a datacentre with a jet engine is perfectly reasonable

Image:
Stop making sense

What distinguished the software revolution from all those that preceded it was instant and potentially infinite reproducibility of digital goods. One unit or a million, the cost remains much the same, and thanks to the internet those ones and zeros can quickly be distributed all over the world.

This has led to a new way of thinking where instantaneous results seem infinitely reasonable, and scale is just a matter of will. Recently it has extended to the financing of AI companies where a million becomes a billion becomes a trillion just like that, and impossible goals like Sam Altman’s aim of adding a gigawatt of compute power a week go almost unchallenged. Just strap a jet engine to your datacentre. job done. Earth’s finite resources? Mine the moon. Nothing is real, but the magic money being chucked around mean that AI is the only game in town.

Spare a thought, then, for the poor IT decision maker trying to navigate this hall of mirrors. Among registrants to our recent IT Leaders Summit, AI - and how to deploy it - was mentioned as far and away their biggest concern/challenge, towering over cost management and skills shortages and dwarfing even cybersecurity in a year when everyone is getting hacked.

The pressure to integrate AI into workflows is intense. But to what end? And what will be the side effects on organisational culture, individuals and society? What about security? Actually, scrap that: will it even work?

There’s a fascinating contrast to be drawn between this intense focus on AI and the money being spent. Our research into IT budgets and their allocations found very little movement in spend on generative AI (the one everyone is talking about) compared with the same period last year. This small survey may not be representative, but it does make you wonder who’s actually buying the stuff. Like many things, it doesn’t add up.

At which point I feel obliged to say I’m not a naysayer. In recent days I’ve heard from a good friend of Computing about an application developed in days by one skilled practitioner and a couple of LLMs which would have taken a team of 10 several months. I also sat down with a thoughtful CTO of a company developing AI solutions and was impressed by how blue-sky thinking is being grounded by engineering nous.

Which puts into perspective the extrapolation from next-word predictor to eliminator of hunger and solver of climate change (issues, incidentally, not unsolvable through a lack of intelligence, rather well-understood conflicts of interest) that’s driving the current craziness. Or maybe we just need to strap another couple of engines to the DC.

CTO Alex Duncan left the British Heart Foundation just as its transformation peaked. She told Tom Allen why, and where she’s going next. Another CTO, Databricks' Dael Williamson, chatted to me about the challenges in getting agentic AI to work for businesses. And to mix things up a little, for our Ctrl-Alt-Lead podcast series Tom spoke to a CDIO, APEM Group’s John Seabourn, podcast series about how AI is already changing the workforce there.