IBM: Think AI Hype

IBM’s annual Think conference was held last week and I have to wonder how much thinking actually happens over there after watching the keynote by CEO Arvind Krishna.

Krishna gave a keynote full of the usual tired bromides about how more technology is the answer to every question you might care to ask. In a complete coincidence, IBM is a technology company that would be only too happy to sell you some.

Yawn.

In a speech that could have been given at any tech conference by any executive from any vendor, the buzzwords du jour were hybrid cloud and artificial intelligence. Hybrid cloud because IBM’s attempts to build a viable public cloud offering failed, and artificial intelligence because Watson wasn’t quite a large enough failure the first time and IBM needs to try again.

IBM is the lucky beneficiary of enterprise inertia (some might say not for the first time) in that hybrid cloud is what existed as soon as cloud became a thing and not everyone embraces new ideas straight away. IBM’s inability to turn itself into a public cloud company, despite concerted efforts to do so, ironically paid off as the tech world realized that not everything was going to go into public clouds.

That was never the real point of cloud, but “cloud is a state of mind, not a place” hasn’t quite sunk in yet. The situation wasn’t helped by AWS’ insistence for years that “thou shalt have no other clouds but me” until huge piles of enterprise cash proved irresistible and now you can say hybrid cloud without security escorting you out of re:Invent.

But I digress.

IBM has one hammer to hit all your problems with and it’s called Red Hat OpenShift. IBM cautions against picking one or more silos, such as a public cloud competitor, and urges you to pick its unified silo instead.

IBM had to buy it, rather than build it itself, I assume because its artificial intelligence systems were designed by Finance. Whatever you want to do, OpenShift is IBM’s answer.

Unless the answer is AI, which IBM is at pains to insist is built on OpenShift, ensuring that the hammer gets a mention here, too.

AI All The Things

As the topic shifted to AI we were subjected to a stream of empty buzzwords about ‘unlocking’ and ‘value’ and ‘productivity’. One could be forgiven for thinking the keynote was written by a Markov chain generator trained on breathlessly positive articles about ChatGPT, which, in a way, it was.

Krishna was keen to start splitting AI into ‘consumer AI’ and ‘enterprise AI’. This is so that IBM can call whatever it doesn’t sell ‘consumer AI’ and thus claim a leadership position in a market it gets to define. A neat trick if it can pull it off.

Krishna claims that one reason for the distinction is that enterprise AI has to be accurate. “You can’t really afford for it to give you incorrect answers,” Krishna said. A statement that is pretty damning about AI in general, and IBM’s attitude to consumers specifically, if you actually pause to think about it for more than a second. Which we don’t.

IBM wants to deploy AI at scale. Presumably not at the scale of consumers, because there are a lot more of them than there are enterprises, but let’s not get bogged down with joined up thinking. We have computers to do that for us now.

Maintenance is for Losers

Krishna quipped that 70% or more of IT people spend their time on operations and maintenance. Apparently that’s a bad thing, and he wants it to be more like 30%. Just think that through for a moment.

Doubling the amount of new stuff created with zero additional maintenance would imply that either the world will fill up with even more stuff, faster. Or, and this is possibly the more disturbing implication, a lot more stuff would need to be destroyed to make room for all the new stuff. What stuff would that be, exactly? IBM doesn’t say.

We would also have to somehow ignore the forces of entropy. That would require us to leave the confines of the physical laws of this universe and travel to whichever reality IBM currently inhabits. Handily, we won’t actually need to do that.

This same claim that 70% effort on maintenance is bad has been made for decades and yet zero progress has been made on reducing it. We know this because the same claim keeps getting made, year after year, keynote after keynote, tedious, buzzword-laden sales pitch after tedious, soul-destroying pitch. Nothing has changed to make this prediction any more credible than the last fifteen times it was made other than the name given to the technology that—this time for sure—will finally rid us of the drudgery of human existence.

Unless you’re the people being paid a pittance to clean the data these glorious AI wonders need to work, that is. Pay no attention to that man behind the curtain.

Replacing the Humans

Despite an earlier claim that engineers will only become more productive—they won’t lose their jobs!—Krishna lets the mask slip when he starts to talk about consumers again. That’s you and me when we’re not at work, by the way.

Consumers are pesky things that ask questions and need to be cared for. Right now, that means employing a lot of humans to do the messy work of maintaining relationships with people. Fear not! IBM will help you to replace all these humans with computers that are better than humans in every way that matters to IBM.

“These agents don’t get tired, they don’t lose their temper, they can work 24/7, they can scale,” says Krishna. He could be describing an IVR system or the Terminator, and I get the impression that it would evoke just as much enthusiasm.

No time is spent wondering why people have these questions in the first place, or how systems could be designed differently to be less confusing. Instead, the solution is to automate bureaucratic bandaids and declare success.

Happily, IBM’s ability to deliver on any of its wild-eyed predictions is suspect, at best. No doubt many IBM customers will waste a lot of human time and effort on doomed attempts to replace all their staff with robots, but it’s the thought that counts.

An AI Mid-life Crisis?

My impression is that IBM has plastered what it already has with as many current buzzwords as possible to hide the true nature of what lies underneath. It smacks of a slightly pathetic need to seem relevant, like a recently divorced dad dressing in his teenage son’s clothes.

IBM seems to hate itself, so it is desperately pretending to be someone, anyone, else.

Which is a shame, because IBM has plenty of genuinely good products that solve actual, real problems for people. It could talk about those, but large sections of the tech industry seem obsessed with the possibilities of the far future rather than more prosaic needs of the here and now. Which is weird, because the customers here have money and want to spend it now. Futurist hype works best in growing economies with low inflation and near-zero interest rates.

IBM offers no pathway from here to there, though it could have, just pretty words about how amazing the future could be. It will arrive as if by magic, overnight, with AI doing all the hard work for us. If we just clap our hands together and believe strongly enough, everything will work out great!

I enjoy an escapist fantasy as much as anyone who isn’t a joyless husk, but unfortunately I have adult responsibilities, too. We need tech companies that understand the more prosaic, human needs of keeping people housed and clothed and watered and fed.

IBM used to be such a company, a company for grownups. Hopefully one day it will come to its senses and decide to be one again.