Remix.run Logo
▲ alex43578 7 hours ago

I don’t get this knee-jerk association of AI = lies or wrong. AI-written content can certainly make mistakes, but so can human content.

The author says “That you're disadvantaged by not using it… Apparently actually knowing things doesn't matter nearly so much as knowing how to prompt a machine to do the thing.”, which is hilarious coming from someone taking an engineering degree.

Just like how CAD and FEA have supplanted hand-drafting and manual calculations, so too will LLMs supplant hand-writing a report.

Newsflash: you will absolutely be disadvantaged by not using tools to advance and accelerate your work. For FEA, you need to understand and validate the results - you don’t get extra credit for doing the math (or in the case of writing, doing the writing) by hand. Use an LLM, understand how to prompt it, validate/read/refine the results.

▲throwaway19268 7 hours ago | parent | next [-]

LLMs were built on the sum of decades and decades of manual human effort and knowledge. What you're talking about is exactly the delegation of thinking the author is talking about. I ask to what end the reliance of AI leads, it's clear LLMs lack the same level of creativity and ingenuity of real humans, and even if they were on par, if we've delegated all the thinking away then who's to say we'll even understand the results. There is a _massive_ difference between using CAD tools to remove the need for manual calculations vs prompting an AI to write an entire report on a subject you have zero understanding of. And I believe you cannot truly build an understanding without at least a moderate amount of effort.

▲alex43578 6 hours ago | parent | next [-]

“It's clear LLMs lack the same level of creativity and ingenuity of real humans” is completely unsupported, particularly in light of LLMs directly generating answers to questions humans have been unable to solve for years.

I’d agree that simply prompting “Write a 10 page report on helicopters” is going to leave them without understanding. But providing the data, refining the output, and adversarially evaluating the output of a computer are all essential skills for a modern engineer. There is no value in effort for effort’s sake. We don’t make students learn everything from first principles and force engineers to develop their own equations just because it would be a “delegation of thinking” to build on the work of others, whether it be from a textbook or LLM.

▲OroPla 5 hours ago | parent | prev | next [-]

Isn't delegating the unfun parts to focus on the fun parts exactly what we wanted from AI?

I don't like writing a ten pages report, nor do I like reading a ten pages report. If it was up to me we would just communicate as tersely as possible everywhere instead of filling things up with a lot of nothing. It's why I rarely read (or write) articles, but read a lot of headlines and comments instead. One gets to the point faster this way.

▲yunwal 6 hours ago | parent | prev [-]

> prompting an AI to write an entire report on a subject you have zero understanding of

It continues to astound me all of the people who insist on turning off their brain just because they’re using one tool and not some other tool. Like, did you turn off your brain when you first used a computer to solve a math problem? You’re still dealing with a prediction machine. It’s a very very powerful prediction machine, but it’s a prediction machine nevertheless. So just… don’t stop thinking when you use it? You literally don’t have to interact with an LLM this way, no one is telling you to have it write your paper for you end to end, you are doing that to yourself.

▲Toutouxc 5 hours ago | parent | prev [-]

> I don’t get this knee-jerk association of AI = lies or wrong. AI-written content can certainly make mistakes, but so can human content.

The issue for me is that I can tell [what I think is] low-quality material from high-quality material based on all kinds of different signals. How the author describes the problem and their motivation to talk about it, whether they suspiciously leave out things that I consider important, whether they're overly reductive or repetitive, whether the order of their arguments makes sense. Those are all valuable signals that immediately tell me whether to spend time and energy incorporating all that information into my mental model.

AI text is all the same, flat, there's no signal. Reading AI text is like watching a movie muted with someone explaining in detail what the characters are talking about. It might seem equivalent to some people, but to others it would feel absolutely insane. It reminds me a bit of those people who say that you can get an "equivalent" machine to a $2k MacBook for $1k, but they're only equivalent if you compare the RAM and CPU frequency and ignore 123 other parameters.