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. …
Close to what?
A cursory look at the Yuki Capital site, it’s a poster child for AI slop. That the AI CEO runs autonomously and promotes code is infuriating nonsense.
I don’t myself have any ability to evaluate what sort of outcomes it is delivering as CEO from what’s on their website. Or from my company’s CEO’s communications either! I also am very averse to judging a CEO’s performance by any short term outcomes, which is a problem for many companies today … a disincentive to take the long term investments view.
I certainly don’t see what is particularly challenging about CEO than other jobs being done by AI, other than that current CEOs are motivated to not reduce the workforce that is them. Making decisions with incomplete information? Dealing with abstract concepts? Synthesizing diverse inputs and making sense of them? These are things AI can do now.
That “AI” will be doing the bidding of human leaders, and they will probably be even worse than they are now. There’s no Wizard of Oz without someone behind the curtain.
I’m starting to suspect your responses in this thread are AI generated.
Are you running some sort of social experiment?
Here’s today’s preview of what “AI civilization of the future will look like”:
Infrastructure engineer: We determined that MSFT has been over-charging us for service X, thousands of dollars, even though we had the service shut down 98% of the time.
Manager: Could we use Claude to tell us what other hosted platform services MSFT is billing us for?
(…pregnant pause…)
Infrastructure engineer: …what do you mean?
Here’s an amusing video of a YouTuber hacking an AI scam call.
Don’t Hang Up On AI Scammers. Do THIS Instead. - YouTube
But it does highlight some of the potential issues of just sort of mindlessly letting AI run things.
That’s not teaching them at the same time. That’s teaching two different students in two private lessons. Now those private lessonsmight be going on simultaneously, on slate screens and separate audio. But that’s not the two students sitting in the same classroom hearing the same lesson together.
Yes, learning is a complex system that gaining information is only a part of.
Consider every class that you took in school from kindergarden through whatever advanced degree you hold. How much actual content of any of the lessons do you remember? How many facts and figures, dates and events, specific bits of lessons do you remember?
But you remember skills you acquired and practiced. Addition and subtraction, multiplication tables, etc. You have a sum of knowledge that you can call upon within your field that far exceeds the lessons that taught them. You probably don’t remember a single lecture, but the information from those lectures got absorbed and integrated into your thoughts and memories and skills.
I have a practical example of multi-stream learning. Back in jr high and high school, classes consisted of copying notes by hand while hearing them. This created multiple channels of learning - seeing, hearing, and processing in order to write the words. Then would come review questions, or homework problems. And I found it relatively easy to remember the material not only for the immediate class but for the subsequent tests later in the year, or the following year, etc.
College is when I hit classes where the lecture notes were pre-printed. We could follow along with our copy and add additional comments, and not be “distracted from something said by trying to keep up”. The trade off was I struggled to retain the material, and had to develop new study skills I hadn’t previously needed.
Just like the prepared handouts are now digital lessons, the multi-stream learning is not happening for today’s kids.
So one observation about social media and social impact comes to mind. I’ve poked around on reddit, and there’s a sub called “am I ugly?” that people post to and ask strangers to answer. There are numerous subs with similar themes about appearance and being rated, etc.
The fascinating observation to me was not this existence. Rather, the large number of submitters that are gen Z saying that they have never been on a date or gotten compliments in person. These for the most part are healthy, fairly attractive young folks - some very attractive. What stands out to me is that while some just might be people fishing for compliments, the bulk of them seem genuine that they don’t get in-person compliments or requests for dates.
Apps have certainly taken over the dating sphere.
Now I personally witness one attractive young woman who works for me get more than her fair share of male attention, most of it unwanted but tolerated. But that so many of that generation are feeling like they don’t get the attention they do want in person leaves me pondering if what I’m seeing is not mere selection bias, but a more widespread phenomenon.
AI is going to have similar unpredictable changes to our social behavior.
With regards to the improvements AI promises in knowledge and ability to give truthful, meaningful answers, I have two examples.
The first, I had to call a helpline to assist a customer and attempt to find out what was going on with their order. The AI phone client on the other end took the customer’s name and correctly identified the order, and told me when it was placed and that it has been delayed until 3 pm. The problem was that it was already 5 pm when I called, and the AI did not know that, nor have a helpful response when informed of that detail.
Second item - my company has implemented a new AI safety observation system using the cameras in the store to look for slip, trip, and fall issues. Since it is new, we are undoubtedly in the learning phase, but right now the system is not very helpful and actively stupid.
For example, the system starts flagging alerts when the afternoon stockers are working, identifying the carts and boxes that are actively being worked as questions if they are safety risks?
The other things it seems to harp on are bits of trash like snips of paper or small decals from produce - flat things the size of a dime. I’m hard pressed to see how that is a safety issue in any way. But I have to interrupt whatever I’m doing because the app alerts on my phone, and I have to verify these are getting responded to or they get escalated up the management chain.
I hope that I’m making a difference in training with my responses, but I wonder how fast the learning curve (not very yet) or if my efforts are being undone by someone else’s responses.
That sounds similar to the concept of “looksmaxing” I’ve been hearing about as well. Which I guess is maximizing your physical appearance to be in the top x% of men who get all the girls.
As I’ve said before, I think these sort of things is a symptom of a broader fragmentation or atomization of society. Basically breaking cohesive groups of people into more isolated individuals who are at the same time connected to a broader more superficial network.
It’s the difference between going to a party and maybe competing with a dozen other guys for a woman’s attention based on a combination of appearance, personality, and general vibe and accepting the natural tradeoffs vs going online as an abstracted digital avatar and competing against the entire world based on am analyzable collection of metrics and KPIs.
Now add to that AI being able to analyze everything about you, find every possible fault or flaw, and then use that to manipulate you into buying more crap or voting for more assholes or whatever.
The “no request for dates” is an interesting phenomenon as I’ve heard several theories that apply to a lack of socialization in general. In addition to the online interactions competing for real world attention, another theory I’ve heard is that the fear of potentially being recorded doing something “cringe” is preventing a lot of in person interaction.
Including hitting yourself in the face with a hammer.
This sounds like a very unhealthy level of neediness to me – the constant need for re-assurance and likes is made worse by AI. Makes the movie “Her” seem like a dramatized documentary.