Are we offloading too much of our thinking to AI? (artfish.ai)

The system does not reward deep understanding. It's too slow.

An analogy that I think helps describe how I feel about it all:

AI is like having cheat codes in a video game. You don't have to try hard and develop a deep understanding of how the game works. You aren't challenged anymore and it doesn't feel like you earned winning the game.

It doesn't matter because nobody cares. Businesses truly do not care. You are a cog, a means to an end. It's only about "winning". Now it's no longer fun to play the game.

I've been a manager for some years. Without sounding too up myself, what made me good at management was the fact that I knew how to do the things I asked people to do. It allowed me to critically evaluate the work, set realistic timelines, and champion the contributions up the chain. I also maintained the ability to perform this work (by design and happenstance sometimes, since we could be understaffed due to sickness etc).

The good managers I saw also had this ability. The managers I didn't enjoy working with / saw struggle were ones who either never had that ability, or had lost it a long time ago.

Note I'm speaking about line management. There's a world of difference between managing a dozen or so individual contributors vs managing senior managers / directors. However, managing LLMs is analogous to line management.

Yeah, I've always tried to train myself to do calculations in my head as much as reasonably possible when learning about mathematical objects, etc. Like when I was learning linear algebra I made myself invert 4x4 matrices in my head. (Pen and paper is also cheating!) Calculators and computers have been better than me at this sort of thing for my entire life, so in some sense this isn't a change?

I am curious thought. If nobody is qualified to answer, how would your team have approached the problem without any AI?

> Very few people actually "think".

Unknowable. And a callous notion.

I don't relate with this at all, are you saying that founders who execute through employees are not creating anything by "themselves", they have no "unique contribution"? Most of us just want to create great products and hopefully get paid for them. And at the end, if I create a great product through this method, I feel equally happy, who wants to grind for a decade over some manual software creation process ?

LLMs are bicycles for the mind, they do not think instead of you

The calculator argument is interesting because I am for sure slower at arithmetic by hand than when I was in gradeschool able to rattle off times tables and long division and what not, due to calculator use. I know plenty of other people who can’t easily do basic math, like multiplying something by 1.5x as a recent example in my life. Their gradeschool self probably would also run laps around them.

> calculators didn't make us dumber

For most people, GPS did not improve sense of direction, spellchecking did not help to write without making mistake, deepl did not help to be better in foreign languages. But replacing a bicycle by a motorcycle forces to acquire new skills without losing any, and we can find many example of symbiosis between "The man and the machine" (Lindbergh wrote a book named "WE"). AI could be something like that, after all it is human knowledge reachable in a conversational and contextual manner.

So AI can be used to learn: "tell me what's wrong in my code, or if it can be improved". I also tend to think that the more we code, the more we give AI valuable piece of knowledge to learn from, the best code it can produce, the less the produced code seems alien. It can be a win/win, all depend on the mindset. I like to code and even if I am skeptical about many aspects of AI I can share the workload with a robot, as an exercise or if the time or budget is constrained.

> is it the prompt you once wrote?

Yes, but also the prompts you give after, and the iterations you continue to do. Art is rarely something you build once and forget about. A lot of that going on in AI no doubt, but the builders who polish their work and strive for perfection are still the builders they were before AI.

You mention the novel writer and yes that person who's great at writing novels will create great art. But the ones that are 80% of the way their, but maybe can't nail the ending, or segue into different plot-lines, can now create great art and have AI get them there. Subconsciously, they will get better at those things by using AI and seeing the result.

I'm always reminded of that I, Robot scene where the scientist says "you must ask the right questions." And that is true now more than ever.

> calculators didn't make us dumber

I think that most of people misunderstand what calculator changed. Calculator didn't replace people doing math, calculator replaced mathematical tables, slide rulers and other already existing devices.

And regarding making people dumber ... Math teachers who saw this change said that there was a clear shift – students started to think less critically. With slide rulers and tables you had to think about answers – significant figures etc – with calculators you don't.

it's the gut bacteria that is left. lol, that hunger, that emotion, that sense of intuition. Without emotions we are just a naturally evolved LLM.

The thing with calculator argument that always gets me. I do math unconsciously all the time. Even something as simple as adjusting a recipe when cooking. I don't need to grab a device to do that, I just do the quick math in my head. And the only way I can do that is because I learned how to do math. And even with a calculator, I still needed to know what to calculate. The argument "we don't need math, we have a calculator" assumes you always get a textbook question that lays it out for you.

Same goes for LLMs. I can use them for programming, and they're very convenient. But I still need to know what to ask it and make sure it stays within the confines of what I want. And without my knowledge I would have no clue if what it's trying do is correct, or safe.

Naturally, this assumes a workflow where you do actually look and modify the output yourself. But I'd argue that any non tech person is inevitably going to hit a wall where they can't debug themselves out of without getting a human involved.

> If a human writes a novel, it's inherently worth more because it's hard-earned and anchored to experiences we share

But the market doesnt reward that. Different consumers different markets, but (using a salad analogy) overall consumers dont care how a salad is made, only whether it tastes good and is cheap

Just to leapfrog off of the calculator notion...

There's a difference between capability and mastery. If you have a calculator, you are capable of doing basic arithmetic (probably). If you can do fairly complex arithmetic in your head, that's mastery. If all you need is capability, it's fair to say that acquiring mastery might be wasted effort. However, if you want to go on to do advanced math, physics, etc., then mastery of basic arithmetic, calculus, algebra, etc. are necessary stepping stones. If you have to go back to calculators, math programs, textbooks, etc. every time for basic things while trying to formulate and manipulate equations, you're going to have a very hard time getting along.

Sure, rely on AI for things that you merely need as capabilities, but recognize that doing so prevents you from developing mastery required to progress to other things. If you don't want to progress to those things, then it's fine. If you do, then you should rethink your approach.

I'm acutely aware that even 6-12 months off the keyboard is brutal to come back from, it's a real issue now because there are pockets of work where the agents can drive the whole thing. If you just autopilot there? The comprehension and skill debt becomes really hard to pay down in as little as a week or two for the really high complexity stuff.

I try hard to keep a serious feature branch going on every project where I'm driving it in emacs or vscode or whatever. `gptel` is great for this because it can tool call for the docs, be a super search engine like Google used to be but you're keeping your reflexes up, keeping the comprehension debt under control.

For me this is almost exactly like being in an EM role: it's so easy to lose track of the details, you have all these other demands on your time, there's pressure to grow headcount, and the costs of letting the skills slip is all but invisible for a while. But before you know it you're rubber stamping stuff you don't really grasp, you're not able to help people who are having trouble in the codebase, it's all bad. And depending on why you got into this business, it can become pretty joyless to be totally out of the loop.

Now every engineer has to navigate this. It's going to be interesting.

> In my mind, it's pretty simple: I'm a human, LLMs are not. If a human writes a novel, it's inherently worth more because it's hard-earned and anchored to experiences we share.

In my mind, it's pretty simple too and not dissimilar to what you say: we've got cognition, LLMs don't. When a human writes a novel, it doesn't look like a novel because the human has been trained by reading every single novel ever written in every single human language out there.

The human doesn't imitate the act of writing a novel: the human just writes a novel.

> But to latch onto the calculator argument: if you outsource adding numbers to a calculator, you're still you. On the flip side, if you use an LLM do most of your thinking, what's left?

Really not much. The IQ of an oyster more or less. But the LLM is not thinking, so it's not possible to "offload your thinking".

But tbh those who believe these LLMs are thinking machines and that it's normal to offload "thinking" to those non-thinking machines probably don't have very deep thoughts anyway.

So nothing of value is lost.

P.S: I happily not one but three AI subscriptions and use LLMs daily to code. Which is probably why I'm such a critic of AI. They're both useful and pathetically bad. They really write crappy code.

> Many LLM maximalists say they use the tech to learn new things, but to what effect?

I would also argue that they use LLMs to learn "about" new things, but not really the thing itself. I, myself, learned a lot about complex SQL queries just recently using LLMs. But I really don't know how to write complex queries myself.

People often confuse reading about a thing with knowing the thing.

Eh, not convinced. Certainly what works for you works for you.

But the idea that “outsourcing” to experts is somehow detrimental is one of the modern tropes I can’t stand. And I’m not sure I believe there’s a material difference between e.g. reading research about raising children versus asking an LLM that is grounded in (and cites) that research.

Maybe I’m a cynic, but I don’t think most of us have a unique contribution to the world. Emphasis on unique.

So why not leverage all resources in the effort to maximize happiness for ourselves and those around us?

>But to latch onto the calculator argument: if you outsource adding numbers to a calculator, you're still you. On the flip side, if you use an LLM do most of your thinking, what's left? We have people here who use LLMs to raise their children, to manage relationships, to design products. So what's your unique contribution to this world - is it the prompt you once wrote? You're standing in front of a token-generating machine, pulling a lever, sometimes receiving gifts. Is that your edge, your unique experience, your purpose in life?

Cognitive offloading seems to be a feature of humans, we do it with spouses and people we work with.

I dont see why offloading some other stuff to an LLM is going to dehumanise someone.

My take is that you learn math before learning how to use a calculator. And can understand what it does and reproduce it (slower) by hand if needed, but everyone is still using it for stuff we don't need to think about.

If you use AI exactly like this I find no issue, but I fear most people are using the calculator without knowing the math and that's the real issue to me.

TBH, the calculator example isn’t an encouraging one. I use a calculator (these days N PP, or spotlight). But by the time I do that I already have an approximate ide of why the answer is, so if it’s way off, or the sign is wrong, I see it immediately.

Same with GPS, though many folks have no idea how to navigate on their own anymore so often don’t notice when they’ve accidentally had a typo or chosen the wrong Springfield.

So I find using an LLM can be useful when I already have an idea of that the problem is.

At their request I reviewed a colleague’s plan for a project. I looked up some of the baseline parameters so I could understand the plan, than I made some comments w/questions about things I didn’t quite understand.

A couple of minutes late for our morning mtg I came in while my colleague was saying that I had found a bunch of things that Claude had not.

This is why they are bad news for new grads who haven’t yet learned the and pitfalls of their fields. And this was a senior person who just outsourced the work to the LLM.

> I'm not good at lifting weights or running, so my thinking is the only thing I have.

Even if you were, the best you could hope for is a career in entertainment ("sports"). The necessity for bodily strength was abolished in the industrial revolution of the 1800s and 1900s. Now the maximalists predictably say we have another "industrial revolution" that makes our intellect redundant. I don't believe that but if it were true, the question what defines the value of a human would become even more pressing, with less leeway as to what the answer might be. H. G. Wells needed The Time Machine to find a world like that.

Still, if enough people believe that LLMs do make our intellect redundant, it will trigger a cultural crisis similar to the one at the previous turn of the century when men struggled with their defining attribute (physical strength) being taken away by machines. The reverse centaur is over 100 years old, even if that name wasn't used then.

This just sounds like the black and white fallacy ad nauseum. Like every sentence is constructed to frame the entire issue around a false binary. Like you assume that if "maximalists" use LLMs more than the average joe, they're not "thinking." Black and white. Either-or. Maximalists are hobbyists. They usually read up on the newest models, try out different ones, maybe even build pcs specifically to run local models. You honestly think they're not thinking as much as you? Also there is absolutely no reason to value "being unique" for its own sake. Unique people aren't walking around trying to avoid tech to max out their uniqueness potential. That's just neuroticism. It does seem to me as though a lot of people have this misplaced faith in humanism and it just feels like copium at this point. Worse yet, it honestly just seems like a way to dunk on those they perceive as conformists. So basically just a high school mall goth. You are not in a position to lecture other people about their purpose. Your brain is a pattern recognition machine just like everyone else. You've grown fond of some patterns you picked up on that emerged during the enlightenment and you seem to be just superficially regurgitating them, not unlike an LLM. Calculators made us dumber in the sense that we no longer have to perform mathematical operations on the fly. That gives us room to do something else. What other people do is up to them. That's what freedom is. You assume the maximalist is just sitting there prompting all day. If they are, whatever. That's their hobby. It’s not much different than playing video games all day. Either way, Im pretty sure their lives are filled with sadness and joy just like you and everyone else.

Having learned math in an accelerated program that banned calculators, my experience has been that they do substantially diminish mathematical learning. They are useful for things like building an intuition about the behavior of functions... but ask a student to manually graph and compare the behavior of some fairly trivial functions ( ln x, 1/x, sqrt x, x^3 ) over the interval [0, 1] and you'll pretty quickly see the difference between kids who are trained to think for themselves and kids who aren't.

The ability to engage with math "critically" is fundamental to being able to use it constructively/creatively.

I think we're seeing the impact of LLMs more broadly in the drastic decline in incoming freshman academic competency levels, and will continue to observe people becoming measurably, meaningfully stupider until we clamp down on / outright ban LLM usage in education.

> On the flip side, if you use an LLM do most of your thinking, what's left?

If your answer is "not much", you gotta step up so that the answer is closer to "more than ever was".

You dug the holes in the ground with a shovel your entire life. Suddenly somebody gifts you a digger, nearly for free. If your reaction to this is "what's left for me now? anyone could just pull levers, no strength or skill needed!" not, "I could dig so much more and deeper holes with this digger!" then you need to chill, reflect, then fight your fear and perverse laziness and learn how to become the best person to pull the levers.

Taking the context of raising children, the ‘unique contribution’ of a lot of parents is traumatising their kids. If it’s helping us make better decisions, I don’t see how that could be worse. I think that translates to all the non creative things you spoke about.