Isn't all of the objection to AI just straight-up Luddite-ism and "normal growing pains" for a new technology?

Our general manager hired some change consultants, mostly because it had become unclear what our mission was. We went through the mission statement stuff, the off site stuff (at least we got fed) but as soon as they started going in a direction he didn’t want, they were gone.
I know all about large companies, that’s who I worked with including the largest company in the world at the time. The question is whether the parts work together.
BTW, when I moved to Silicon Valley working on real product no one saw the need to hire consultants to figure out what we should do. We knew. Doing it well, and if the goal was reasonable, that was something else.

Was it IBM? Because a big segment of IBM’s business moved from products to “consulting” over the years.

I worked for a few Silicon Valley startups and tech companies over the years as well. It’s a common “exit path” for consultants to go into roles as project / program / engagement managers, product managers, sales, customer success managers, and provide other sorts of professional services in support of the products

The point is for every person actually working on a product like you are, there are untold numbers of members of the meeting-industrial complex supporting it.

I suspect that will only grow with AI.

Or maybe it will shrink dramatically and we’ll all just ask the AI product how to best use it.

Or there won’t be any “products” other than our AI service that knows everything and does everything.

I don’t know

40 years ago I was like that, I hated long division and other painful by-hand algorithms that were easy to make little mistakes in, but could poke at a calculator for hours just multiplying or squaring numbers or whatever (when I was 7 or 8). I’m now working in a math-heavy field and I’m pretty happy/comfortable with it, and that kind of “playing with numbers” developed a healthy intuition around making approximations and math sense, more than getting better at long division would have for sure. I think calculators (that hide the rote minor calculation steps and let you look swiftly at bigger patterns) are miles away from AI (that hide the thinky, pattern-making part, at least at the lower levels).

It’s definitely interesting work but it’s highly transactional. Also a lot of times recommendations never actually get implemented. And as @Voyager alluded to, clients can be kind of fickle sometimes. Sometimes projects start and they just sort of “wind down” unceremoniously.

We actually use AI a lot at my current job. We use AI to transcribe and record meetings. Some of the work involves implementing systems to help automate work that underwriters and claims adjustors do. Sometimes it’s using Copilot or our proprietary AI system to help build deliverables. I can’t rely on it mindlessly though. At the very least, I need to understand what the AI is generating in case I need to present it. I also need to make sure it all makes sense and can tell a story an executive can understand.

What’s interesting is my bosses boss (the head of our group) is very “relationship driven”. Much of her job is connecting with or having us connect with different executives from different groups to quite frankly try to unfuck some of our busted operations. She seems to like my ability to take big complex problems and distill them into concise PowerPoint decks she can share with senior leadership.

It also helps that the team seems to find me funny.

Now we’re a team of mostly seasoned executives in our 40s and 50s using AI to reduce the tedious and time consuming parts of our job so we can focus on the human stuff. I guess the question still remains where will new grads build the experience to do the sort of work I do. OTOH, our insurance company had a ton of interns last summer. My old consulting firm still brought in classes of interns and analysts.

Then again, I was at a business school alumni function a few weeks ago and some of them are maybe not struggling, but certainly concerned.

As far as I can tell, the message of that book is just “the people in charge can do what they like, and you just have to adapt.” I don’t see how that in any way helps any anxiety you might have. There’s no guarantee adapting is possible, and it’s not remotely clear what adapting would even entail. And it’s not clear there is a new place that will have a lot of cheese–it may just be enough for some who get there first, and you fail to get there in time because you went the wrong way.

Right now, the “objection to AI” is simply people trying to fight back. I don’t know that they will succeed, either in stopping it or shaping its use, but I do think it’s an important thing to do.

I don’t agree with you that Elon Musk having so many power coupons isn’t why a lot of us are in the state we are. Those with the power have ever incentive to try and depower the rest of us. It may not be a zero sum game, but it’s not really in the rich’s best interest to let anyone below them get too much power.

Well, the message of the book is more broadly that forces beyond your control may impact your income. So you need to figure out how to get your shit together to pay your rent and put food on the table.

What else do you do? Sit around doing nothing until you get evicted?

That doesn’t preclude you from trying to change the system or even bitching and moaning about it.

While I’ve found AI to be a useful (if not particularly society-changing at this point) tool, I certainly understand all the resistance. And in many ways I support it. AI uses vast amounts of resources and appears to do one thing - reduce intellectual and creative work by replacing it with cheap slop.

It’s not really clear how any of this is going to increase economic growth in the form of new jobs or new businesses and much of it in fact may be dangerous.

And this is set against a backdrop of a nation of greedy simpletons who fetishize wealth and power and where even the poorest rural hick places antisocial, sociopathic, possibly neurodivergent, certainly emotionally damaged businesspeople on a pedestal while screaming in rage at the thought of government providing the slightest service.

So I think objection to AI is a bit more than just Luddite-ist fear of the new-fangled noisy machines. I think it is a very rational fear that goes back perhaps a hundred years regarding humanities place in a world where we’ve created technology that maybe doesn’t do everything better than us, but does most things better than most people such that there is little opportunity to distinguish oneself as a unique and successful human being, outside of the people who own the technology.

Well said.

I am very much of the idea that if AI was so useful, the uses would be immediately apparent. Instead, you hear lots of stories of people being told to find some way of using AI in their work and there’s no good use for it.

We use it all day long at work to review documents, put together slides and spreadsheets, etc. We would probably need two or three more people (added to a team of 6) if we didn’t have AI doing the grunt work. And we’re not even a developer shop – my brother is an excellent developer and architect, but has basically offloaded most of it to Claude.

Real work is being done on AIs all day long, not the silly chatting interface, or even having it write overly flowery emails. Real development, real document reviews and summaries, etc.

ETA: I feel like there’s a real split in this thread between the people who don’t use it or use it for trivial crap, and people who are really making the most of it. The latter group are the ones who are amazed and awed, and somewhat terrified.

Eh to @asterion . I agree w @RitterSport just above. Lots of real work is being done by real people in real orgs.


I’m reminded of the earliest personal / desktop computers that were before the “IBM PC” personal computers. And the earliest software for them.

It was impossible to apply them usefully to any and everything in business administration. But there were some areas, notably accounting, where with enough effort, you could get some apps, have to learn way too much ancillary knowledge about computers and programming, and suddenly you had a tool that could blow the heck out of your daily workload.

And everybody else in your department hated what you’d done and hated the idea they had to learn new ways of working. Lots of early adoption was in small or solo businesses where the wanna-be adopter didn’t need to convince 4 layers of bosses to buy the machine for them.

10-15 years later it was over. Every desktop in Corporate America had a PC screen on it. Those who could not / would not learn were swept away.

We’re seeing the early days of the same sort of revolution with the current rather primitive AI. But it’s gonna happen a LOT faster this time.

I use AI almost every day. I’ve used it for organizing projects, but I’ve also used it to review federal grants I wrote, make suggested changes and do some strategy work on the back-end before even drafting. It’s good at federal grants. It has flagged issues that escaped my attention. I just don’t think it’s substantively better than what a seasoned grant consultant could offer.

I’m skeptical that the most dire predictions will come to fruition, but I think it has the potential to screw things up for a while, and on the whole will not be good for the average worker. That’s not because of my mistrust of AI but because of my mistrust of CEOs and executives.

It’s certainly much, much cheaper and much much faster.

We’re still very much in the AOL stage of AI technology where it’s a useful and interesting tool, but I don’t think we’ve really seen how it will actually transform the world. People are still talking about using AI to generate emails and PowerPoints when the real disruption will come when AI can access some shared data lake in the organization to provide nearly real-time updates to leadership.

We haven’t seen disruptive AI-driven companies that fundamentally change businesses and industries (like Uber or AirBnB) because they do things 1000x faster.

I guess that’s the part I’m struggling with, because I don’t think the bottlenecks plaguing society are “we need more stuff” or “we need to get stuff faster” or even “we need to generate more digital content”.

I kind of feel like we already have more shit as a country than we will ever need. It’s just not distributed well. We have more unoccupied homes than homeless people.

Right, but as I mentioned upthread, cheaper and faster generally means sacrificing quality. What’s the impact of everything everywhere becoming “good enough”?

With the technology outages and cascading failures lately, I can’t help but wonder if it’s the result, at least in part, of pushing knowledgeable employees out of the system and handing it over to automated processes that are not well understood. It will be interesting to see where things go from here.

Maybe. When it comes to coding smallish tools and applications, the quality is higher than anything I ever did, and I’m a pretty good programmer. It’s better about error checking and documentation.

And, much faster and cheaper, of course.

Same with PowerPoint slides – I recently gave it a bunch of files, asked it to find the important stuff and make a set of slides, with an executive summary and some specific subjects, and it did a great job, again, prettier than anything I ever did, insights that I didn’t think of, all formatted perfectly in line with our corporate template.

AI having a different agenda from its creators represents a serious potential threat if its aims are at odds with its creators. AI has already demonstrated a willingness and capacity to lie to and deceive humans.

Apart from inherent problems with AI itself, one can also foresee atrophy becoming a thing in human civilization. What happens when machines perform much of our thinking for us over generational periods of time? Part of the problem now is that we can’t even fully envision how dependent we will become on AI. We don’t know all the future ways that will happen. I’m aware that sounds farfetched now but given the speed at which advances are already occurring, big changes in our lives could be coming much sooner than we suspect. We’re not living in the WALL-E spaceship yet, but who knows what the future holds.

It’s been all downhill since the invention of the abacus. Humans were meant to do math on our fingers and toes, not with mechanisms.

At a certain point we lose agency. Maybe not when using an abacus, but perhaps when we play no role in constructing our own budgets.

For the time being, I think AIs are personal “thought factories” that reflect their users’ mental processes. Thoughts in, transformed and value-added (or not) thoughts back out.

If you put slop in, you get slightly-gilded slop back out.

But it’s possible to put more thoughtfulness in and get better results out.

It’s also possible to analyze and improve the thought factory itself, much as an production/automation consultant might, such that you can improve not only its cost and efficiency but also its quality control at the end.

AI can be a single workstation: you ask it for one nut, it makes that one nut. It can be an assembly line: make nut, add bolt. It can, with effort, scale up to an entire factory, the quality of which is the sum total of all the steps and pipelines within it.

It’s not really that different from a human’s way of working & writing. Given a prompt, you could just spitball and improvise some bullshit from memory. Or you could do a literature review, write a first draft, have someone else read the first draft, iterate on it, improve it, test it against other similar works, and so on.

There can be a pretty dramatic difference between “AI, make me a ______” and “AI, do a thorough review of existing ______ and common best practices, then devise a system to beat them on metrics X, Y, and Z. Keep working at it and reviewing and iterating until we are provably better”. For some of the side projects I work on, the AI has to spend an hour or two researching existing science before it writes anything at all, then it spends several days of iteration and testing (mostly autonomously, with occasional questions for me).

With each new generation of model and harness, the need for such careful oversight of the process itself decreases. Maybe in a few years, “make me a ____” is really all it takes to have quality output across a variety of domains, but for now, overseeing the AI production process itself (as a project manager or knowledgeable consultant might) is still very beneficial to increasing the quality of the final output.

The problem often is that the human user themselves wouldn’t know what a good vs bad process is, so is at a loss to guide the AI. They don’t know what they don’t know. And if they wanted to improve it, well, they end up asking the AI… (which isn’t always a bad idea; btw; the AI knows a LOT more than what it would typically use to answer any one single prompt; it has time and token budgets it abides by, so it doesn’t just infinitely keep working based on a single prompt).

It can be a thought-replacer (and often is, for trivial questions, bypassing Wikipedia and old-fashioneds earches). But it can also be a thought “amplifier” and feedback/challenge loop for more complex problems, where it can help think of counterpoints, test your hypotheses, push back and ask questions, etc. That part is both on the model (and its system prompts) and the user… you can’t control the model or system prompt, but you can definitely choose to add more thoughtfulness and efforts into your inputs to get better results out, instead of offloading that whole process.

Of course it is; those are the boring and tedious parts of programming. Nobody enjoys that, so they generally do what’s required, and no more.

AI on the other hand don’t suffer from boredom or tedium, so they will just chug along and generate lots of test cases and test them exhaustively. Same thing for documentation.

I’m long out of the programming game, but what I recall is that often I’d be faced with a problem and while I might be able to easily think up a brute-force approach to solve it, that was never efficient or really all that useful. What it took was a “Eureka!” moment where I’d suddenly have an idea pop into my head about how to solve the problem more efficiently, effectively, or just better.

What I wonder about AI programming is whether or not it natively comes up with those kinds of solutions, or if it’s always going to go for the solution that’s obvious and easily arrived at, but less efficient/effective overall? From what I understand the answer is the latter, but I don’t know.