When I am working on an email or I have to write something that I'm not quite sure how to phrase, I notice that "I could just throw it in Chat" flashes across my mental marquee.
I have yet to actually try this, because I don't believe that an LLM has any intuition and also I don't want my communications to sound like a greasy machine. Yet, the thought is there. For a fragment of a second, I look to AI to save me.
Perhaps it is because I see so many people talking about using LLMs for this or for that, and it is clear that the wider world believes sincerely, deeply, childishly that LLMs can save them -- they can write the email. A draft at least. Probably come up with an outline or some stock phrases for how people say these things.
Why do I think of AI saving me if I don't actually believe it can save me? What I think is happening is that I don't like doing difficult things. This isn't new. I don't want to have to focus hard and think through how my words will be received. This is especially true when I'm writing something that is beyond my comfort zone -- where I'm unsure about what it is that I'm saying or asking.
So, I look to AI, this great god in the sky, to save me.
I don't believe it can save me. But, like a reverse of every religion on Earth, it impresses with daily miracles and doesn't seem to need my faith at all -- just my monthly subscription.
(and all my data)
For now, my attitude is one of resistance. I'm not going to take the easy path and have Chat write my complicated emails. It's possible I am become a noble savage, uncivilized, in the back eddy of progress.
I'm on the resistance side too and I publish less posts because of this. It's less enjoyable when choosing between either writing inferior posts or falling back to the monotone, soulless autocorrect.
The third option where we attempt to get the best of both worlds is 2-3x labor and time intensive versus not using the LLM and I am too swamped right now to do it. HitL means you are the bottleneck and this can be rather disillusioning.
But do you find the brief flash of "I could just have AI solve this problem" is changing how you think? The presence of this new option changes the experience of confronting the problem.
I'm trying to figure out how the presence of a "Chat can solve this" solution changes how I think through my own solution to a piece of work.
hmm I guess I should first disclose that I don't truly believe that "AI" (I read this as "an LLM") solves anything. Remember that it only does what it is trained to do + your prompt + give or take some fuzziness from randomizers. So the answer is no, I'll try to explain:
What could be the case is that there is some sort of SOP for the thing you're doing that either you don't know about or find too bothersome to follow, but the bot was trained on executing it. Even though bots get nerf'd all the time now (especially Claude), with enough budget and patience you can probably instruct it to do what you need, with some success rate between 50-80% [1]. It's always "bots could maybe solve this" but never a definite yes; it always remains to be seen what comes out.
What did change this past year is how I approach input. LLMs have a better flagging rate than I do because they don't get tired or bored. So I allow bots to pass me slop, then I read the slop, then I assess findings I find worth pursuing one by one. So discovery has gotten some steroids, but you need to be able to assess what it says, which is a good reason to never c&p bot slop to anyone other than yourself (because you asked for it - no one else did and if they wanna, they can ask themselves.)
Bottom line, I see the dilemma as: if you don't care about quality or your reputation, do whatever. If however you do care about these things, there is something to keep in mind: if you have deep understanding of what you're doing, bot output will be inferior to your own. If you don't know what you're doing, bot output will be inferior to those that do know, but you won't notice that.
I really have not succeeded in getting a consistent >80% success rate, in my #1 instruction field, security pre-assessments. I jokingly say that this is a skills issue on my end whenever people challenge me on that, but it's not really... it's more that a lot of anti-patterns are trained into the bots that most people, including the AI labs, would call "good enough" but in reality it's all relatively low standards. I also fear (and hope) that generically trained bots will always be inferior to specialists of the meat or silicon kind (and no, skill files do not remove training correlations.) ↩
Solution may have been too strong a term on my part.
Here's an example: I want to write an AMA announcement for someone we have coming this week. It doesn't take long because I have a format I use for all of them and that gives the post a lot of structure. However, I like to say something interesting about each guest, kind of like a hook.
Often when pull up notepad to write this kind of post, there's this fleeting feeling that I can just have Chat generate the post. It will likely be good enough for social media. Whereas if I do it, it will probably take 5 minutes because I really want to try to make it a strong hook (often I fail at this, but the desire is the point here).
In this sense, an LLM can probably solve my problem by taking far less time and producing something that is good enough.
So, the thing I'm talking about is this: how does it change how I think and ability of my mind that I've always got this little "easy button" I can tap? -- even if I don't tap it, I resent the way it intrudes on my thought process.
I think what I am trying to say above that this feeling is deceptive. So while I understand the temptation, it's just something you have to guard yourself against, because it's the wrong use of the tool. Especially because:
Honestly, I think you've got the right idea here. You don't let an LLM write something about someone else (unless it's a vibe song, which would be fun) because doing that the "easy" way out is directly showing how much you value the person that you're hosting. When you host someone, I think it's best to spend the time. More so than any post about the latest fake L2 (though there too it is good to just write your own thing.)
I'd try to shift perception from "LLMs deliver a product" to "LLMs deliver input". I'll give you an example from my end:
I owe k00b a plan. If I would have generated it or even parts of it, it would have been done last month, because unlike me, LLMs don't get fatigue or writers block. I do. But this stuff is important to get right because I've been designing a privacy feature and the plan covers how to implement it. One does not fuck these up, and prior experience tells me that LLMs fuck up all the time. So, it's slow, but it will be good. The way I use LLMs is to fact-check me, to test my plan for errors. To check for shit I missed and to audit end-to-end correctness. But not, absolutely not, to write the end product.
This is an excellent framing and I'm going to use it a lot I think. It short circuits my complaint and perhaps if I spent more time thinking about LLMs providing inputs, it would change this weird mental desire to be lazy.
But here's my next question: imagine we live in a world where LLMs don't fuck things up. How does this change how we think?
This may just be so much navel gazing though. And not particularly useful.
Note: fucking up is subjective!
Current architectural issues aside, this is probably attainable when you train (not write prompts and skill files for) ScoresbysAssistant™. And I'll have OptisLittleHelper™.
At that moment it will not change your mindset, because you will own it. It will be your tool, not someone else's rotten magic. You'll be you but then you'll truly be able to do 10x productivity.
LLMs are just an amplifier(at least for now) so you should use it to make your output 100x better/faster/bigger.
Allow LLMs to make you work harder on this you trully love, not to solve tasks you dont want.
So my diagnosis is that if you feel they are making you lazy you are not working on the right things.
(could be very wrong :P)
I mostly agree with you, but what I'm trying to identify is this strange change where I know always have someone (an LLM) to ask.
It's like the difference between working out in a gym to stay fit and living a lifestyle that requires lots of physical movement.
It used to be that we had to figure things out or impose upon another person to help us figure them out. Now there's always an LLM.
I'm trying to think about how this availability changes how I think.
Got it.
The availability is indeed very novel, plus the availability and willingless of LLMs.
However this is not completely new. People have been outsourcing small task to asistants or other peopel that work for them for some time :)
reminder & caution: Artificial Interactive Computation is still a baby; who's ur daddy?
I have no doubt that LLMs will solve all their tonal issues at some point in the near future. The complaints we lodge now about sounding like AI are not so different than the complaints about pictures with six fingers -- soon to be a thing of the past.
However, this won't change the weird dynamic on which I'm trying to put my finger: how does the having an ever-present easy solution to all thinking tasks change the way we think?
there really are no shortcuts, ever - if the right way is the hard way, then so be it - get going! choosing the easy way is an opportunity cost for proper hard training; the hard way is very hard cuz we have been retarded for so long; once the first step of the hard way has been transcended, the next step will obviously be even harder, requiring further training;
case in point: @Darthcoin's Citadel; the audience may think that this first stage was crazy difficult, cuz they have never attempted anything like this; the next stage is already out of their reach, mentally & physically;
one shud proceed up the spiral staircase, without clearly seeing what's beyond the next turn;
still tempted to ask the artificial computation tool to come up with a message without contemplating the alphabeth, the sounds, & the effects of vibrations? ngmi; #1283918
That "citadel" was just learning how to improve from my own mistakes. I did a lot of mistakes and I know it.
That's how you became sovereign: govern yourself.
i finally got to work an excavator alongside working with various-sized shovels as well as my bare hands with the rocks & dirt & weeds - what i have confirmed for myself is that without understanding how a small patch of dirt behaves & knowing the ultimate objective... terraforming on a large scale is often counterproductive, requiring huge efforts to correct later - best start small, steady, and slow;
working on correcting ur mistakes is the self shadow-work;
This also happens to me. I usually end up just not writing it.
I think differently about this. When it comes to work emails, I quickly dump whatever’s going on in my
mind and turn to ChatGPT to refine it to a more formal tone.
I have been called out when I tried to inject my funky personality into previous emails. So I don’t think people at work will appreciate me not toeing the line. Also I think if I finish writing emails faster, I would have more energy for the real work that is educating my students
Interesting. In the US culture in which I swim, many funky flourishes are accepted. Perhaps East-coast banker culture is different, but my impression here is that an overly formal tone is usually seen as more risky than informality.
But I don't think I was trying to solve a tone issue. More that some of the hard work of writing is choosing the organization of the ideas and chat bots offer an endless supply of speedy solutions to this. Knowing this can change how we write.
Do you worry that using ChatGPT to refine the tone of your emails will result in you losing somewhat your own tone?
not really because I don’t have to write professional emails all that often in the first place. I think the greater issue with regard to my situation is that as a teacher, I feel compelled to write compositions that conform to the writing conventions flavored by the system — which may be at odds with what I consider good writing. case in point: my countrymen seem to like adverbs, which I think weakens one’s writing. so to a certain extent, I have sold my individuality to the system haha. but I don’t actually mind since I have little aspirations of writing my own fiction and stuff. I’m happy enough writing compos that will serve my students as well as the random pieces to share with the good folks at SN
Try DeepL (https://www.deepl.com/en/write)
It makes ai texts sound more natural and human like.
I'm not so worried about how the chatbot sounds. I am worried that now that there is an easy button, it is present in every attempt to write, and that changes how we write.
I feel ya.
I have the same temptation. I think it comes from feeling like, I'm stuck now; getting something decent out, even if tonally boring, is better than staying stuck.
Which may be true, in the moment. But what does it do to us when we consistently turn to this crutch? It likely short circuits our own capacity to think deeply, or at the very least to express ourselves with precision and depth.
better than staying stuck for sure. We are so conscious of "wasting time" and being stuck feels like "wasting time" so the temptation is to solve the problem with mediocre output, because it's better than the crime of time wasting.
I'm still worried, though, that spending some time being stuck is somewhat important.
The name is sympatic.
The urge to use AI and the urge to resist it seem to coexist in a lot of us.
You might just be the last of a dying breed as everyone is hopping on the AI trend. It's no longer the hype marketing many believed it to me. Real world solutions are getting solved by it but I still appreciate your resistance.
To me it's like a drug, something to use in time of need and not to get addicted to it.
Writing forces you to confront exactly what you think, which is why it’s painful. AI doesn’t save us from the work; it just saves us from the friction of starting. The real magic happens when you edit the machine's draft into something human.