The world will be consumed by AI.

Once upon a time only the brightest (and / or richest) went to college. So a college degree becomes a proxy for clever.

Now since college graduates get the good jobs, the way to give everyone a good job is to give everyone a degree.

And since most people are only interested in the job, not the learning that underpins the degree, well, you get a bunch of students that care only for the pass mark and the certificate at the end.

When people are only there to play the game, then you can't expect them to learn.

However, while 90% will miss the opportunity right there in front of them, 10% will grab it and suck the marrow. If you are in college I recommend you take advantage of the chance to interact with the knowledge on offer. College may be offered to all, but only a lucky few see the gold on offer, and really learn.

That's the thing about the game. It's not just about the final score. There's so much more on offer.

Learning is not just a function of aptitude and/or effort. Interest is a huge factor as well, and even for a single person, what they find interesting changes over time.

I don't think it's really possible to have a large cohort of people pass thru a liberal arts education, with everyone learning the same stuff at the same time, and have a majority of them "suck the marrow" out of the opportunity.

For us the curriculum was the start of the learning, not the end. We'd get a weekly assignment that could be done in an afternoon. Most of the class did the assignments, and that was enough.

There was a small group of us that lived (pretty much) in the lab. We'd take the assignment and run with it, for days, nights, spare periods, whatever. That 10 line assignment? We turned it into 1000 lines every week.

For example the class on sorting might specify a specific algorithm. We'd do all of them. Compete against each other to make the fastest one. Compare one dataset to another. Investigate data distributions. You know, suck the marrow.

(Our professors would also swing by the lab from time to time to see how things were going, drop the odd hint, or prod the bear in a direction and so on. And this is all still undergrad.

I can imagine a History major doing the same. Researching beyond the curriculum. Going down rabbit holes.

My point is though is that you're right. You need to be interested. You need to have this compulsion. You can't tell a person "go, learn". All you can do is offer the environment, sit back, and see who grabs the opportunity.

I get that you cant imagine this playing out. To those interested only in the degree, it's unimaginable. And no, as long as burning-desire is not on the entry requirements, it most certainly will not be the majority.

In truth the lab resources eoild never have coped if the majority did what we did.

By 'liberal arts' I meant the common 4 year, non-vocational education. My major was CS too, but well over half of the time was spent on other subjects.

> I get that you cant imagine this playing out. To those interested only in the degree, it's unimaginable

I can easily imagine what you describe playing out. I just wouldn't call it 'sucking the marrow' (unless you were equally avid in all your classes, which time likely would not permit).

But as you allude to in your last point, the system isn't really designed for that. It's nice when it does effectively support the few who have developed the interest, and have extra time to devote to it, as it did for you.

I'd rather see systems that were designed for it though.

This is because that is what companies care about. It's not a proxy for cleverness or intelligence - it's a box to check.

If you accept that the degree/job relationship is the start, not end, of the reason for being there, then you see other things too.

There are opportunities around the student which are for them, not for their degree, not for their job. There are things you can learn, and never be graded. There are toys to play with you'll never see again. There are whole departments of living experts happy to answer questions.

For example, (this is pre google) I wrote a program and so needed to understand international copyright. I could have gone to the library and read about it. Instead I went to the law faculty, knocked on the door, and found their professor who specialized in intellectual property.

Since the program I wrote was in the medical space, I went to the medical campus, to the medical research library, and found tomes that listed researchers who might benefit. I basically learned about marketing.

If all you care about is the company job, then all you'll see is the degree.

Then fail to actually learn anything and apply for jobs and try to cheat the interviewers using the same AI that helped them graduate. I fear that LLMs have already fostered the first batch of developers who cannot function without it. I don't even mind that you use an LLM for parts of your job, but you need to be able to function without it. Not all data is allowed to go into an AI prompt, some problems aren't solvable with the LLMs and you're not building your own skills if you rely on generated code/configuration for the simpler issues.

That is, the job of a professional programmer includes having produced code that they understand the behavior of. Otherwise you’ve failed to do your due diligence.

If people are using LLMs to generate code, and then actually doing the work of understanding how that code works… that’s fine! Who cares!

If people are just vibe coding and pushing the results to customers without understanding it—they are wildly unethical and irresponsible. (People have been doing this for decades, they didn’t have the AI to optimize the situation, but they managed to do it by copy-pasting from stack overflow).

I have met maybe two people who truly understood the behaviour of their code and both employed formal methods. Everyone else, including myself, are at varying levels of confusion.

Playing the contrarian here, but I'm from a batch of developers that can't function without a compiler, and I'm at 10% of what I can do without an IDE and static analysis.

Sure, there's a huge jump from a line editor like `ed` to a screen editor like `vi` or `emacs`, but from there on, it was diminishing returns really (a good debugger was usually the biggest benefit next) — I've also had the "pleasure" of having to use `echo`, `cat` and `sed` to edit complex code in a restricted, embedded environment, and while it made iterations slower, not that much more slower than if I had a full IDE at my disposal.

In general, if I am in a good mood (and thus not annoyed at having to do so many things "manually"), I am probably only 20% slower than with my fully configured IDE at coding things up, which translates to less than 5% of slow down on actually delivering the thing I am working on.

Same with more advanced editors and IDEs. They help with tediousness, which can hinders insight, but does not help it if you do not have the foundation.

What would you consider AI in it?

A compiler translates _what you have already implemented_ into another computer runnable language. There is an actual grammar that defines the rules. It does not generate new business logic or assumptions. You have already done the work and taken all the decisions that needed critical thought, it's just being translated _instruction by instruction_. (btw you should check how compilers work, it's fun)

Using an LLM is more akin to copying from Stackoverflow than using a compiler/transpiler.

In the same way, I see org charts that put developers above AI managers, which are above AI developers. This is just smoke. You can't have LLMs generating thousands of lines of code independently. Unless you want a dumpster fire very quickly...

(Yes, these are people with developer jobs, often at "serious" companies.)

Maybe you mean people who are bad at interviews? Or people whose job isn't actually programming? Or maybe "lots" means "at least one"? Or maybe they can strictly speaking do fizzbuzz, but are "in any case bad programmers"? If your claim is true, what do these people do all day (or, let's say, did before LLMs were a thing...)?

> Maybe you mean people who are bad at interviews?

No, the opposite. These developers learn the relevant buzzwords and can string them together convincingly, but fail to actually understand what they're regurgitating. (Very similar to an LLM, actually.)

E.g., these people will throw words like "Dunder method" around with great confidence, but then will completely melt down for fifteen minutes if a function argument has the same name as a module.

When on the job these people just copy-paste existing code from the "serious company" monorepo all day, every day. They call it "teamwork".

I’ve met some really terrible programmers, and some programmers who freeze during interviews.

Predictably they end up with some people on the range from "can't code at all" to "newbie coder without talent"

I don't. I think the world is falling into two camps with these tools and models.

> I now circle back to my main point: I have never seen any form of create generative model output (be that image, text, audio, or video) which I would rather see than the original prompt. The resulting output has less substance than the prompt and lacks any human vision in its creation. The whole point of making creative work is to share one’s own experience

Strong disagree with Clayton's conclusion.

We just made this with AI, and I'm pretty sure you don't want to see the raw inputs unless you're a creator:

https://www.youtube.com/watch?v=H4NFXGMuwpY

I think the world will be segregated into two types of AI user:

- Those that use the AI as a complete end-to-end tool

- Those that leverage the AI as tool for their own creativity and workflows, that use it to enhance the work they already do

The latter is absolutely a great use case for AI.

I am not a creator but I am interested in generative AI capabilities and their limits, and I even suffered through the entire video which tries to be funny, but really isn't (and it'd be easier to skim through as a script than the full video).

So even in this case, I would be more interested in the prompt than in this video.

"Tall man, armor that is robotic and mechanical in appearance, NFL logo on chest, blue legs".,

And so on, embedded in node wiring diagrams to fiddly configs and specialized models for bespoke purposes, "camera" movements, etc.

Seeing this non-compelling prompt would tell me right off the bat that I wouldn't be interested in the video either.

The video is not exactly great, IMO.

Because those who recruit based on the degree aren't worth more than those who get a degree by using LLMs.

Maybe it will force a big change in the way students are graded. Maybe, after they have handed in their essay, the teacher should just have a discussion about it, to see how much they actually absorbed from the topic.

Or not, and LLMs will just make everything worse. That's more likely IMO.

Please, dear god, do not cut the funding of that program preventing the spread of these northwards.

First viable airplane shell is anticipated to hit the market in 2250.

/s

Yeah, I stopped all contact that same day. That project never got launched.

I've gone to some of these just out of curiosity and it seems like people on the hunt for suckers. Having been thru this myself and seen the drama play out with friends from back in the dot-com days to now --

CO-FOUNDING IS LIKE MARRIAGE You cannot just match to someone. There has to be a low-stakes dating period, an engagement period, and then "marriage". You have to be able to walk away early on w/o extensive entanglement. The best co-founder is someone you've known for a long time in various semi-tense scenarios, where you can evaluate their ethics under pressure. This ideally means having worked together or done many projects together.

Buying a used car presupposes the ability to service it.

I bet they have more than half the original range, so I don't have to consider a new battery at all.

https://www.reuters.com/investigates/special-report/tesla-ba...

Idle draw/phantom drain of 5-10% a day seems like a common complaint with Teslas as well:

https://news.ycombinator.com/item?id=43314781

I personally don't want to be left holding the bag on a used battery since I like the option traveling long distance without stopping, but obviously it's your choice.

Yes, but I don't think it's enough to affect this comparison.

> I personally don't want to be left holding the bag on a used battery since I like the option traveling long distance without stopping, but obviously it's your choice.

The point is we're comparing against a car that has 150 miles of range brand new, 120 or less by the time you'd replace its battery. A Tesla battery at 60% is still better.

> Idle draw/phantom drain of 5-10% a day seems like a common complaint with Teslas as wel

That's a more serious problem, among many others. I'm not advocating one way or the other, just focusing on range.

Like anything used, or new, what you'll get is random. In general, Tesla seems to have their batteries sorted out compared to other manufacturers, but you can be unlucky.

The accuracy of range varies a bit, like most cars. If you have your foot in it, you'll do worse, and if you drive efficiently you'll do better.

But I'd want that car to be under $15k. That car doesn't exist, at least not in the US, so I'm still on an old ICE.

My current car is worth at most $5k, and I spend maybe $500/yr on gas.

My wife uses it for commuting and it's our standard "run to town" car. We have an ICE vehicle for trips.

Absolutely would buy again.

Consider for example a very long road trip, such as a drive from Los Angeles to New York which is around 2800 miles. Suppose Alice has a 200 mile range EV and Bob has a 300 mile range EV. Assume their cars both charge at 300 miles/hour.

What effects will the range difference have on their trips?

• If DC EV charging stations are too far apart the trip might not be feasible for Alice.

• Assuming DC EV stations are not too far apart for Alice, then she will have to stop to charge more often.

• Assuming both start fully charged, and during the trip they drive until 10% and then charge to 80% then Alice will drive 180 miles until her first recharge, and then recharge every 140 miles. Bob will drive 270 miles until his first recharge, and then recharge every 210 miles.

• Alice will have to recharge 20 times. Bob will have to recharge 13 times. That works out to be a stop about every 2 hours of driving for Alice and a stop every 3 hours of driving for Bob.

OK, that is a lot more stops for Alice. But lets look at the time spent stopped rather than the number of stops.

• Alice first stops with 2620 miles left to go. Her car recharges at 300 miles/hour, which means she needs 8h 44m of total charge time.

• Bob first stops with 2530 miles left to go. His car recharging at 300 miles/hour needs a total charge time of 8h 26m.

Let's assume the each charge stop adds 5 minutes of delay beyond the actual charging time. Alice stops 7 times more than Bob so has 35 more minutes of delay. Add that to the 18 minutes more Alice spends actually charging and we have Bob gets to New York 53 minutes ahead of Alice.

Suppose Alice's car could charge faster than Bob's? Say 500 miles/hour? Then she would only spend 5h 15m charging. That's 3h 11m less charging time than Bob. She'd still have 35 minutes of delay from more frequent charging but even with that she'd beat Bob to New York by 2h 36m.

This suggests that for long road trips charging rate may be more more important than range, as long as range is sufficient to let you reach DC charging stations.

I say this as an EV owner btw

The SUV is in all likelihood quite uncomfortable, as it has to fit into the surrounding pickup.

A small affordable truck works great as a commuter and picking up supplies from a hardware store for home improvement.

Not if it has two seats and you want to carry more than two people.

>picking up supplies from a hardware store for home improvement.

I don't remember where I heard it. But it was something like "most people buy cars for reasons least likely to happen". If you aren't using your bed daily you bought the wrong car. Buy a trailer if you need to go to a hardware store once a month.

Then you don't want a small truck.

> If you aren't using your bed daily you bought the wrong car.

Plenty of cars have a second row of seats that are not used daily too. I guess they bought the wrong car as well.

Plus not a lot of cars have trailer hitches...

Yes, that was my point.

>Plenty of cars have a second row of seats that are not used daily too. I guess they bought the wrong car as well.

Cars with only two seats are in almost all cases more expensive. To buy a pickup you pay extra, to get 4 seats you don't.

What you personally need I have no idea nor was my reply pertaining to that.

Gain, people do not buy cars they need. They specifically buy cars for situations they will never find themselves in.

Your scenario here makes even less sense because most cars do not come with a trailer hitch (trucks on other hand...), you're buying another thing (which lowers that this is supposed to be an "affordable" EV), you have to renew registration for the trailer, and have a place on top of your vehicle to store it.

A truck bed is nice to have sometimes. You can quickly throw things in it (like plywood, a new large flat screen, bicycles, fishing gear, a dead deer, etc). You don't have to hookup a trailer or fiddle with wiggling things between doors and so forth.

Such as commuting to & from work. But I imagine you believe that is the least likely thing to ever happen to a person. Driving to and from work.

Additionally range dictates how often you have to charge. This does not matter if you can easily charge at home, but when you have to do it at public infrastructure it is obviously a hassle, which is increased with lack of range.

GP is pointing out that anything less than 300 is not practical, and they're not wrong. 150 miles, in particular, is just too low to be used for anything other than a basic commuter vehicle. It's useless on highways (try driving I-70 through KS in this thing) and if you change job sites more than once, you're likely to run out of range in one work day. Traveling between cities is going to be unreliable if you're not able to stop and charge at your destination. Live in Colorado Springs and want to go to Denver? You probably have to stop and charge before coming home.

If you have other reqs by all means, pay up. But don’t force the cost on everyone else.

That's 2x the average daily distance driven, 4x if you do it back and forth. This is more like a cheap utility truck for day to day things, not a thing you want to travel in for hours at 65mph

we're also talking in raw miles, straight line, no hills, decent weather. I'm in a hilly city that gets to -30C on the regular, and these factors could turn 150 miles in to 75 pretty fast.