When you lose the ability to write, you also lose some of your ability to think • Hacker News Board

Re: When you lose the ability to write, you also lose some of your ability to think

#561

Post by proamdev123 » Fri, May 12, 2023, 9:05 PM UTC

schwartzworld wrote: post #549

barrkel wrote: post #35

Earlier quoted context omitted.

The tools are at their most useful when they're replacing Stack Overflow searches for common tasks. An example I had last night: how do you split a pdf into separate pages on the command line, and how do you put them back together again (I replaced one of the pages with a scanned signed version). This is a basic task, a beginner's task in the world of pdf munging, but I didn't know the answer off the top of my head.…

This is exactly what I don't understand about it. I copied your question directly from your comment into google and the top-ten results were all sufficient to answer the question. The nature of common tasks is that the answers are also common, so where's the value in paying for a service that regurgitates them to you?

Time, specificity, and interaction.

1. Time — for me to search Google and then read anywhere from 1-10 links to get an answer typically takes anywhere from 5 - 30 minutes. The average is probably 10-15 minutes. But using ChatGPT takes less than 1 minute.

2. Specificity — ChatGPT gives me answers to more specific questions, with no fluff text to read through, no ads, no redirects, no links that appear to answer my specific question but actually don’t when I click on them and read them. I get a specific answer to a specific question.

3. Interaction — If I have follow-up questions and/or need clarification or alternative options/solutions I can simply ask ChatGPT and it answers. I often find that getting answers to followup questions via web search takes even longer than the original search.

These all make it worth it for me. And that still doesn’t touch on all the other use cases that search engines don’t do at all.

Re: When you lose the ability to write, you also lose some of your ability to think

#562

Post by proamdev123 » Fri, May 12, 2023, 9:15 PM UTC

digging wrote: post #389

barrkel wrote: post #35

Earlier quoted context omitted.

The tools are at their most useful when they're replacing Stack Overflow searches for common tasks. An example I had last night: how do you split a pdf into separate pages on the command line, and how do you put them back together again (I replaced one of the pages with a scanned signed version). This is a basic task, a beginner's task in the world of pdf munging, but I didn't know the answer off the top of my head.…

> The tools are at their most useful when they're replacing Stack Overflow searches for common tasks. Interesting, because I still find it faster/easier to do a quick DDG search, with a click on the first relevant SO post, and scrolling straight to the top answer. Asking the same question to ChatGPT means waiting for its output, which may be objectively a few ms faster but creates the perception of slowness ( = irrit…

I often find that the top Stack Overflow answer doesn’t adequately answer my question, so it’s not faster for me.

For example, sometimes the top answer is long and requires a few minutes of reading, sometimes there are corrections in the replies that need to be read, and sometimes I don’t fully understand it and therefore need to read multiple replies.

This can easily take up 5 - 15 minutes, versus 1-5 minutes for ChatGPT.

Additionally, I find value in NOT having to keep myself from getting distracted by the tons of other highly interesting things on Stack Overflow. It can be nice to have that focused, text-only, distraction-free dialogue that ChatGPT provides.

Re: When you lose the ability to write, you also lose some of your ability to think

#563

Post by dctoedt » Fri, May 12, 2023, 9:46 PM UTC

hnfong wrote: post #557

dctoedt wrote: post #494

Earlier quoted context omitted.

> The lawyer. Needs walls of text. Lawyer and law professor here: No, no , NO ! Wall-of-words text is a sign of a L.O.A.D. — a Lazy Or Arrogant Drafter. I teach contract drafting to 3Ls and stress The Two Great Rules of Readability: 1. SSSSP: Short, Simple, Single-Subject Paragraphs. https://toedtclassnotes.site44.com/Notes-on-Contract-Draftin... 2. BLUF: Bottom Line Up Front. https://toedtclassnotes.site44.com/Notes…

(Former law student here...) Definitely agree with your approach to drafting, but I think you could be overreacting a bit here. Even the shortest contracts might seem like a "wall of text" to the lay person, and short paragraphs stringed together is a wall of text. At any rate the preferred format in most legal contexts is still "`wall' of text" (whether big or small wall, as opposed to diagrams, spreadsheets, or con…

> short paragraphs stringed together is a wall of text.

"White space is your friend" is another rule I preach to my students. Some 40 years of dealing with contracts has persuaded me that short paragraphs with white space in between them are FAR easier to review and negotiate — which means they get signed sooner, and clients vastly prefer that.

> as opposed to diagrams, spreadsheets, or conversations

For several years there have been people banging the drum for contracts to include more icons, flow charts, etc., for improved readability. So far, that idea hasn't gotten a lot of traction — IMHO that's because most lawyers barely even know how to use Microsoft Word competently and would have no idea how to create a flow chart or meaningful icon, so they default to writing pure text (which is also the quickest and thus cheapest path to getting a draft onto someone else's desk).

About the best I've been able to come up with in this regard, as something that can be done with Word, is to create tables of, e.g., action items, with columns such as "Event | Action | Who | When" — basically, an Intel-style "AR list" [0], or in software, a CASE statement.

[0] http://www.nathanzeldes.com/wp-content/uploads/2014/01/Effec...

Re: When you lose the ability to write, you also lose some of your ability to think

#564

Post by noduerme » Fri, May 12, 2023, 11:27 PM UTC

nerpderp82 wrote: post #560

noduerme wrote: post #482

Earlier quoted context omitted.

Which is worse, having no debate partner, editor, or quiz master? Or a debate partner who insists they're right when they're wrong, an editor who injects false references and made-up citations, and a quiz master whose accuracy is essentially random?

All those things are already true. Debate isn't about being right, it is about winning a debate. It is literally rhetorical combat. A quiz master that is wrong, how delightful! A chatbot is the multiple choice test you can argue with! If those things are a problem, we have already failed. Because if your current education system is built upon taking things at face value, the harm is not in the shoveled knowledge but…

Yes but "bullshit detection" is a skill honed over time by not only reading many sources, but by learning to read skeptically, analyzing and comparing sources for bias and factual credibility, and adjusting your certainty level on a subject as you go along. No teacher is accurate 100% of the time; lots of teachers spew bullshit when they don't know the answer to something; practically everyone laces their facts with opinions. But you can turn to other primary sources for facts, secondary sources for opinions, and the shape of some general consensus will emerge.

The problem is if there's just one teacher tutoring, one quiz master who points to sources it controls or simply invents as its "proof". My concern is that AI will be trusted more than a single book or a single teacher, precisely because it claims to neutrally encompass a consensus already obtained through all those sources. [It's a tertiary source masquerading as one offering primary and secondary sources, without the ability to differentiate or trust one source more than another].

This is exactly why the frequent comparison to calculators is so misguided: Because calculators are always right. To whatever degree LLM proponents believe LLMs are simply useful tools like calculators, that will be the same degree to which they fail to check for bullshit.

Re: When you lose the ability to write, you also lose some of your ability to think

#565

Post by astrange » Sat, May 13, 2023, 1:08 AM UTC

felixg3 wrote: post #524

loup-vaillant wrote: post #471

Earlier quoted context omitted.

> "make" money (i.e. create value) See, the harmful confusion right there. Making money and creating value are two separate things. There are people who create value without receiving any money in return. And there are people who receive an obscene amount of money while creating very little value — and sometimes destroying more value than they create. Understanding that the two are separate is the only way you can qu…

You’re absolutely right. David Graeber’s “Bullshit Jobs” is a good take on this if one wants to start reading into it.

It seems to be that "bullshit jobs" isn't actually a good position for a leftist to take. You're basically saying workers don't do anything useful.

Even a more thought out form doesn't do anything class analysis didn't already do.

Re: When you lose the ability to write, you also lose some of your ability to think

#566

Post by plumeria » Sat, May 13, 2023, 1:46 AM UTC

hammyhavoc wrote: post #27

rcme wrote: post #14

Sometimes I wonder if I'm using the same ChatGPT as everyone else. I pay for GPT4 access in ChatGPT, and it is a really terrible writer. I also pay for Copilot, and I think a can count the number of times it has been helpful with one hand. I'm trying to get on this hype train, but I'm left with an empty feeling every time I try to seriously spend some time with these new tools.

How impressed or unimpressed you are by it is a direct reflection of yourself and what you are trying to achieve with it. I find zero value in it. Others are very impressed by it, but I'm not very impressed by what they hold up as an example of its output. Sometimes I can only laugh when it's horribly wrong, but they herald it as amazing. The whole thing reminds me of the blockchain hype train. Still using and loving…

Do you have a link to the repo? How does it compare to axcell (which is very domain-specific) [0].

[0] https://github.com/paperswithcode/axcell

Re: When you lose the ability to write, you also lose some of your ability to think

#567

Post by dilippkumar » Sat, May 13, 2023, 1:49 AM UTC

rcme wrote: post #14

Sometimes I wonder if I'm using the same ChatGPT as everyone else. I pay for GPT4 access in ChatGPT, and it is a really terrible writer. I also pay for Copilot, and I think a can count the number of times it has been helpful with one hand. I'm trying to get on this hype train, but I'm left with an empty feeling every time I try to seriously spend some time with these new tools.

I wonder if it’s time to start seriously looking at why some people have completely embraced copilot/chat-gpt and others can not get it to work at all.

I’d love to see which of the following might have any correlations with reported usefulness of chatgpt/copilot:

1. How articulate a person is.

2. How much of an expert they are in what they are trying to do with the tool.

3. How successful they are at giving instructions to a human to perform the same task.

4. How much experience they have managing/coaching junior devs/interne/newbies.

5. How much experience someone has at decomposing a problem into smaller parts and identifying the simple parts and complicated parts

There are huge consequences to either answer to the question: “Is using an AI tool effectively a coachable skill?”. I’m sure someone has already looked into this or is looking into this - if it turns out to be a coachable skill, and we can identify what those are, there’s a lot of money to be made in bringing a coursework to market.

Re: When you lose the ability to write, you also lose some of your ability to think

#568

Post by noisycarlos » Sat, May 13, 2023, 4:59 AM UTC

rcme wrote: post #14

Sometimes I wonder if I'm using the same ChatGPT as everyone else. I pay for GPT4 access in ChatGPT, and it is a really terrible writer. I also pay for Copilot, and I think a can count the number of times it has been helpful with one hand. I'm trying to get on this hype train, but I'm left with an empty feeling every time I try to seriously spend some time with these new tools.

Same here! I can feel like the Emperor's clothes sometimes.

Re: When you lose the ability to write, you also lose some of your ability to think

#569

Post by agalunar » Sat, May 13, 2023, 6:28 AM UTC

hnfong wrote: post #555

agalunar wrote: post #520

Earlier quoted context omitted.

It's completely plausible that a nation with a wealth of natural resources but an economic system that treats people terribly could win a series of wars and establish its economic system as the dominant one in the world, and then use its power to maintain status quo. A marketplace of ideas does not necessarily lead to the best outcomes, or even good outcomes. Suppose the marketplace of ideas is an unregulated marketp…

> It's completely plausible that a nation with a wealth of natural resources but an economic system that treats people terribly could win a series of wars and establish its economic system as the dominant one in the world, and then use its power to maintain status quo. FWIW, Adam Smith has argued that this is unlikely. The world seems to prove him correct. That said I do disagree with the way GP's argues the point. W…

I was trying to be careful by writing "treats people terribly" rather than "treats its people terribly", having imperialist nations in mind, or "treats all people terribly", having class-divided societies in mind.

I didn't realize that Adam Smith had argued on this point, though! Thank you for letting me know.

Re: When you lose the ability to write, you also lose some of your ability to think

#570

Post by nazka » Sat, May 13, 2023, 10:26 AM UTC

loup-vaillant wrote: post #298

noobermin wrote: post #261

Earlier quoted context omitted.

The thing is this will rob a whole generation of that learning. To struggle to learn to write, code, and all that is how you become a professional. A lot of people will never develop these skills, and the chasm between the talented and untalented will only grow.

I recall a study whose result said that beginners improve faster when they're told the solutions right away, and experts improve faster when they're given time to figure things on their own. Struggling as a learning tool is important, but there's a good chance the fastest path to expertise is to start with examples of how to do it, and delay that struggle a bit. What you don't want is to keep the crutch forever and n…

I agree but also I will also argue that the beginners that got all the right answers at the beginning may learn faster in the first steps but will struggle a lot faster later on at the expert level because:

- they wouldn’t have yielded all the knowledge of the little kirks, know how, edge cases, etc… they now have to know (and the Pareto rule 20% vs 80%),

- created an intuition from all the hours spent just struggling to understand what is wrong or find a solution and little by little honing their intuition from that hard work they would have done by themselves for themselves (compared to having ChatGPT, a teacher, or an expert telling them most of time what to do),

- and getting a will to find a solution by themselves even after hours looking at a shell prompt (or logs or whatever else) that is taunting them and will tell them nothing.