Subscribe
About
  • Home
  • /
  • CIO Zone
  • /
  • OpEd: It's not you, AI conversations are really hard right now

OpEd: It's not you, AI conversations are really hard right now

Phillip de Wet
By Phillip de Wet, ITWeb contributor
Johannesburg, 09 Jun 2025
ITWeb contributor Phillip de Wet.
ITWeb contributor Phillip de Wet.

This time last year, Apple announced Apple Intelligence was rolling out across its product lineup, almost exactly as Microsoft was launching its Copilot+ PCs with ludicrous hardware performance to support AI everywhere.

Last week, the console I can't help thinking of as the Nintendo Switch 2: AI edition, went on sale to rapturous response, though gamers won't see the benefits of their expensive hardware in too many direct ways just yet.

It is notable when a new area of focus burns through the consumer space like that, quickly filtering down to an entertainment hardware platform that is not regularly updated. But that's not where the answer lies.

Last week, Amazon announced a dedicated group to build an AI framework for robotics, with talk of general-purpose robots to lift and carry heterogeneous loads autonomously.

It's extremely notable when one of the most ruthlessly efficient logistics companies in the world floats that kind of application for agentic AI, but that's still not the reason, I don't think.

The problem, I reckon, lies in stuff such as the survey that found just how deeply the automotive sector is delving into all kinds of AI applications. Not of the in-car entertainment kind, not the cute and fully stuff you put in front of end-users, but for predictive maintenance, and to enhance factory floor optimisation.

That, right there, is why it is now nearly impossible to have a conversation about AI outside the expert sphere.

Carmakers are up there with brewers and miners in how rooted they are to the real world. Automotive executives who lose sight of the fact that the core of the job is bolting pieces of metal together, they don’t last long. Carmakers are jealous of their R&D budgets, protective of their manufacturing systems, and almost everywhere in the world subject to powerful unions of skilled workers you can’t replace with scabs if they turn on you.

We've reached an important maturity milestone when those people – people not even remotely susceptible to vapourware – have near-term timelines for rolling out AI systems across big swathes of their production stack.

Yet while AI is getting real, tangibly so, there is this utter madness on the dealmaking side.

The speed of the AI revolution has been driven in part by VCs desperate not to miss out on the next big thing, but it has left behind the more staid part of the finance sector, both those who manage money and the people who occupy board seats exactly because they are older and cooler heads.

When it is your job to be on the lookout for the next dotcom crash or the next stupidly expensive stupid fad the youngsters are clamouring about, then “AI” has to raise your hackles.

And those doubters are right – there’s some pretty fishy stuff going on under the cover of AI. When Elon Musk's xAI manages to give itself an $80 billion book value while ascribing a value of $33 billion to Elon Musk's X platform in something vaguely resembling an acquisition, as it did at the end of March, then AI starts to seem like an excuse for exactly the kind of creative accounting that should trigger scepticism.

When OpenAI effectively pays $6.5 billion – well over R100 billion! – for a one-year-old “creative collective” with one legendary name and never talks about what it is actually buying, scepticism turns into alarm bells.

Three weeks since we learnt of that Jony Ive / OpenAI partnership, it still crops up in the vast majority of conversations about AI fatigue and AI hype. Sometimes because true believers want to convey their belief that it will herald the next paradigm-altering device which, you know, sure, maybe, who knows. Mostly, though, the talk is about how some company with “love” in its name ran a heist on some numbskull VCs.

Those who follow the sector a little more closely know that OpenAI also apparently waved $3 billion in front of coding assistant Windsurf, one of several hundred such start-ups that can be wiped out as quickly as they sprang up when Anthropic builds a proper coding system on top of its Claude engine, which most of those start-ups use at their core.

Between these two strands of development, the stupid money on the one side and the practical applications on the other, it seems that every conversation about AI with people in business is now a minefield.

One side is watching the potential benefits mount for whatever sector they’re in, and they are over-optimistic to the point of ignoring the cost of culture change and the shortage of skills and consultants and compute.

The other side is ready to dismiss the whole thing as a scam, because parts of it are.

Most people in the IT space understand that two things can be true at the same time, because they’ve seen the hype cycle play out in one way or the other, and they understand the difficulty in integrating even a genuine magic wand.

The customers those IT people are talking to, that’s a different story.

I don’t know what to do about it, other than wait it out, but I can offer this solace in the meanwhile: it’s not just you. Things are just a lot crazy right now.

Share