I will start worrying when AI takes the job of a tech CEO. I figure if that’s not possible, then neither is most of the doom saying.
Related thread.
Research indicates that people generally think that everyone else’s job is in danger of AI but their job is special and can only be done by a human. That’s because people have a very dim view about what other people do and underestimate the complexity of others’ work. And I know this must apply to business leaders as well.
Chesterton’s Fence Don’t tear something down until you figure out why it was built in the first place.
I would say that’s more of an operational failure than a bad management decision. In my line of work, we try to reduce “shadow IT” as much as possible for the very reason you describe. It creates a huge operational and security risk having these little band aid fixes that actually become a load bearing structure holding the roof up.
My observation is that many execs don’t “manage” anything at all. They have endless meetings where they endlessly discuss things. They have people create management reports but don’t really ever look at them or make any decisions based the (usually incomplete and inaccurate) data. A lot of them are just “middle men” who just pass the word up to their superiors.
I sort of look at it in a similar manner to how an old feudal system operated. Or really any socioeconomic system or hierarchy. They all sort of evolve out of a need to motivate and direct lots of people working on complex tasks in systems where no one can possibly understand how the entire thing works. A lot of these systems are part of what my boss calls “corporate theater”. Often the report itself isn’t what isn’t important. It’s the discipline of forcing people to think about what they have to get done and the perception that there are layers of people who might come in at any time and ask where your report is so they can poke holes in it.
So what does this have to do with AI?
Well, If you had a system that is connected to all these disparate databases and other systems and was reliable and accurate, that would significantly change the necessary management structure in most organizations. In theory it could simply tell you “a web site certificate is coming due” or “that contract has these gaps in it”
Someone in management would have to know that there are such a thing as certificates, or the people working on them would have to document it in a way to get it in the database, I’m not too confident about any of these things.
I was on a committee looking at our performance review process, and we even asked people to give their opinions. The number one issue was that managers went into a meeting with a list of employee accomplishments that the employee never saw. Then they’d get a bad rating for not doing something they did. We fixed this by forcing managers to get employees to sign off on the input to performance review - not that they agreed, just that they saw it.
So I’m less than confident that people “managing” AIs are going to do better.
Not that bosses are the only ones at fault. Before I retired I documented my software extensively. I even put in jokes. I severely doubt anyone ever read it, and it fell apart. I wonder if stuff controlled by AI models that is supposed to last years and years will work when they move to new models. I’ve never heard that brought up as an issue, because things are so new.
Whoever manages your companies web sites should absolutely know this. Not that they always do.
I think the operative work is “performance”. The more systems automate the work, the less you are evaluating the people on what they actually do.
I think that’s the dream driving all this AI investment.
Most managers I’ve worked with are almost completely clueless and lack no real skills besides ass-kissing and presenting themselves as “management”. Not that they are stupid per se. But the system seems to have evolved such that big companies have created these hierarchies of MBA types to oversee this big self-perpetuating machine. But if they need to change something, they call someone like me (or an army of Accenture consultants or offshore tech resources) to fumble along for months or years trying to reverse engineer how this big clusterfuck operates.
And what’s interesting is no one scrutinizes my time even though I’m usually the most expensive person on the team. They micromanage the people doing the actual work and if they aren’t working fast enough they add more layers of people like me to oversee them!
So think the dream is all these senior leaders can tell their “AI wrangler” to check the status of website licenses or insurances LOAs or FINRA certs or whatever administrata (if the AI doesn’t take care of it automatically) or run go to market scenarios or whatever.
After my negative experience losing all that legacy info I have created a treasure trove of documents to aid whoever has to figure this shit out in my absence.
Will it be used? If it’s ever necessary, I won’t be around to find out.
My old department has a lunch every few months. At one my old boss (before he retired) asked me if I could fix up the system again. I said sure, but you’d have to pay me. A lot. That took care of that.
It’s kind of like getting to hang around after you’re dead.
I worked in tech in places where MBAs were useless and more or less unknown. In fact, at Bell Labs they had a good tuition refund program, but MBA programs were excluded from it.
I was a first level manager, which was 95% technical. I filled in as second level manager for a short time, and I found it awful. No fun stuff, just meetings and budgets and politics. My observation was that you don’t have to have an MBA to be an incompetent manager.
Some people beat them, too!
No, but it helps.
It is getting close.
And
the startup plans to acquire a small B2B SaaS or e-commerce company for as much as $1 million and attempt to run it with AI as the CEO. …
… Skyfall has set its sights on what it believes is a far more difficult challenge: a virtual CEO. Unlike software engineering, which he characterizes as largely logical and deterministic, running a business demands abstract reasoning, long-term planning and high-stakes decisions made with incomplete information.
“A lot of startups are building marketing agents or sales agents, which in our belief is very narrow AI implementation — rather than tackling the broader challenge of organizational decision-making,” Pasupalak says. …