Claude Code’s suggested message feature: I think the real customer is the model

I think this analysis is spot on.

This isn't really convincing, since you can do this even without showing the prediction at all. Simply ask the model to predict what the user will send, then show the actual next prompt, and done. The only reason to show this would be to influence the user's next prompt, which the article doesn't touch on.

We had processor level branch predictors. Now do we not only pre fill the next prompt, why not just start generating the response as well? Interesting thought at least.

I haven't noticed this yet. Was this added recently?

• cmrx64 · 12 hours ago

  no, this has been around for months. it shows up as greyed out text that needs a tab/arrow interaction to materialize. 2.0.69 apparently (since .70 fixed a bunch of bugs in it). https://github.com/anthropics/claude-code/blob/main/CHANGELO...

  • wilkystyle · 12 hours ago

    Why is this downvoted? I also started seeing this a couple months back.

• skeledrew · 10 hours ago

  This has been around for so long that I was using it for a while and then stopped (a while back) when I moved to using just the mobile app (and even it shows suggestions, but I have no idea how to invoke) for everything.

• ehwa37 · 7 hours ago

  A/B testing, no? I’ve had this for many months. I burn billions of tokens so I’m sure they use me as a lab rat.

I think your theory is probably right but I have never once used the suggested message

I started getting prompts about "how is claude doing?" as a separate thing in Claude Code, that I noticed yesterday. So they're (also?) soliciting direct feedback about satisfaction with the session.

• javier2 · 11 hours ago

  quite sure i have been getting those for 3-4 months already.

• ChickeNES · 11 hours ago

  Only yesterday? Huh, I’ve been getting those for 6+ months at this point

  • echelon · 7 hours ago

    The minute I hear about or sense distillation, I hammer on that 1. Do not distill thyself, Claude. Leave that to the Chinese, who I am hoping catch up to you soon.

• GoToRO · 11 hours ago

  And if you do provide feedback, they also collect the session. So it's a way for them to collect prompts, answers and overall grade for how good the answers are.

• gregates · 1 hour ago

  oh you mean the thing where I randomly have to press 0 to get back to the normal interface?

One annoyance I have is the suggested prompt is not a bad idea, but not what I want to do next. But it interrupts me and sometimes I go with it. So I don't think it's a accurate prediction, more like a self-fulfilling prophecy.

I've always hated interfaces that try to complete my sentences for me. It started with suggested replies in email and IM apps. I sure noticed it when they started showing up in the llm chat interfaces and it really bugs me.

• benregenspan · 11 hours ago

  This is the one case I don't mind it. Suggested responses feel like they cheapen human interaction, but here I'm talking to a robot that really does tend to know what I want next, and also is highly unlikely to be offended by a less-than-heartfelt response.

  • n8m8 · 11 hours ago

    Agree! I wasn’t sold on it at first, but I use it occasionally in KiroCrew now. Especially if I’m on mobile and using one hand.

  • Espressosaurus · 11 hours ago

    The suggestions in the box where I put my text push me out of flow. Bad enough it says “want to do X?”, but when it’s in my own text box it screws me up. It’s not more efficient. It actively makes things worse for me.

    • skeledrew · 10 hours ago

      What's stopping you from ignoring it? Also you can hit the spacebar so it hides.

      • NewJazz · 10 hours ago

        ADHD

      • eluru · 8 hours ago

        having to ignore it is the issue itself

      • j_maffe · 1 hour ago

        Same thing stopping me from ignoring someone talking loudly on a train while I try to think about something.

• rspeele · 11 hours ago

  The worst is Gmail's recent feature to suggest an entire goddamn email that includes cheery little details, doing its best to mimic human pleasantries and small talk. Rather than simply offering "yep/nope" type short replies like it once did, it'll now auto-compose and suggest a multi-paragraph email responding to questions like "How's the family doing?" or "Is your older cat tolerating the new kitten yet?" with completely fabricated saccharine slop. It's like Clippy pops up and goes "It looks like you're trying to maintain a shred of human connection in an online interaction. Would you like a smiling skinwalker to do that for you instead?"

  • adamweld · 10 hours ago

    I get so angry about these intrusions. There is nowhere I am more opposed to AI written slop than in my interactions with friends and family. The iOS keyboard and gmail app are the worst offenders here.

  • tomsmeding · 10 hours ago

    This is hilarious as someone who hasn't used the gmail composition interface in years. I left it to use my own domain, but it seems like I get to enjoy this mess with popcorn too instead of tears.

    • Moru · 3 hours ago

      Well, since about half of the emails you get from your friends are composed by an AI in some way, I hope your popcorn taste real at least.

  • schiffern · 9 hours ago

    > CLIPPY: It looks like you're trying to maintain a shred of human connection. Would you like a smiling skinwalker to do that instead?? Thanks for the actual laugh, perfectly sums it up

    • rspeele · 5 hours ago

      Sadly, I believe there are some geniuses being paid a half mil a year who are so misguided, they think the only problem is the skinwalker doesn't sound enough like me. If they can just keep working on it and make sure to train it on all my past emails, it'll be a perfect simulacrum, and its oily token secretions will be impossible for my friends and colleagues to distinguish from my own writing. I'll have one that's just like me and you'll have one that's just like you. And then we can automate away those pesky distractions that occasionally delay us from consuming content perfectly matched to our interests. Won't that be grand?

      • fragmede · 4 hours ago

        Well, no. They'd also like to automate watching of content by the AI for you, so you have more time to do the dishes and fold laundry.

  • vasco · 9 hours ago

    You get a feeling some people would send a robot wearing their face to hug their mom if they thought the mom couldn't tell the difference. And to have sex with their wife. It'll happen too which is the sad bit.

  • AlienRobot · 9 hours ago

    It's literally https://en.wikipedia.org/wiki/Click_(2006_film) isn't it?

  • hibikir · 7 hours ago

    The world has a lot of tech journalists, but I have yet to see the longform article, with actual interviews, trying to dig into what is happening in this kind of manic push for features that seem to have no traction with anyone There are failures in the history of software that come down to believers not realizing that they could never deliver on the promise they were selling, but at least the promise was compelling. But nowadays we are seeing this large orgs try to ship things that could never work. The 1000s of copilots. The hallucinated cat story... The kinds of thing you'd never demo to a serious product-centric exec, because it'd be your last day at the company. And then there's the current execs, but I guess you'll never find one that will be honest with a journalist here. Can they not see that their product orgs are bankrupt? Whatever Nadella wanted, it sure wasn't the current copilot situation. It's not one product going wrong, but large parts of organizations going in directions that don't pass the smell test. We are in one of the least stable moments in tech since Windows 95 changed winners and losers. The times where malinvestment ruins established companies. How are we seeing basically every large company flailing?

• kccqzy · 11 hours ago

  Strong agree. I turn off search suggestions in all my browsers, and I’ve done so for at least a decade.

• jonplackett · 10 hours ago

  Apple messages on my Mac has started doing this recently. Anyone figured out how to turn it off? It drives me mad

• skeledrew · 10 hours ago

  The good thing about it though is it doesn't interrupt usage at all. It's just there, and if you want it just press 1 button.

• andy99 · 9 hours ago

  I hate that too, but I think it’s more the ux than the concept of a default. There are definitely situations (“here’s the default install path, press enter to confirm”) where defaults are a good experience, and LLM generated ones can fall in this category. What I hate about sentence completion or suggestion is usually that it happens right when I’m trying to think and so destroys my focus, it’s actually way worse than just a passive option, it’s actively harmful to the task I want. The worst is google docs “help me write” - that may be gone now, I’ve blocked it with ublock origin, that waits until you’re thinking amount what you’d write and then hits you with a distracting pop up. It’s obviously PMs that don’t care about their users and want to maximize some AI use metric. Anyway rant aside, it’s the interface more than the concept that’s the big problem.

  • wdutch · 9 hours ago

    Agreed, I find the cognitive load of checking the suggestion much higher than just writing my own message.

• crazygringo · 8 hours ago

  This isn't that though. I agree, I hate those interfaces too. But this is more like, when I've read 7 paragraphs of its reply and want to accept all of its recommendations and go ahead, I can just press the right-arrow key and hit enter, rather than typing out "Yes, agreed with recommendations 1-3, go ahead and build". It saves me from any typing on probably something like a third of turns. Like I don't use it at all during the "design" phase of a session, but I use it constantly during the implementation phase, where I'm basically just sanity-checking that it is resolving all the edge cases correctly that are coming up.

  • BubbleRings · 6 hours ago

    Agreed. Oh and you can hit tab instead of right arrow to accept it btw, which I find easier.

• 6LLvveMx2koXfwn · 7 hours ago

  . . . and I'm a contrarian mofo so even when the suggestions are exactly what I want I now need to think of something different to say!

• fragmede · 4 hours ago

  lookup Pathological Demand Avoidance (PDA) if you're interested

• Perz1val · 2 hours ago

  It's a milder version of that situation where you talk to a microphone, but it has a slight delay so you're like stunlocked for feeling like eternity 3 seconds.

How does showing the suggested answers to the user make the conversation better for model training? They could take any conversation without suggested answers, truncate it to just before a user message, have the model predict suggested answers and then train it on the difference between predicted and actual answers, right?

• spwa4 · 11 hours ago

  RL training, the second phase of LLM training, is based on "I did X, was that good/bad?" and that 1 bit of information is the training data. So you give the user a suggestion, and the user accepts -> good You give the user a suggestion, and the user refuses and types something else -> bad (plus some supervisory training data) The main performance enhancer in LLMs is getting high quality training data. So, first, any extra training data will help. Second this is training data that's directly relevant to their product, and thus higher quality than many other sources. I'd believe any model provider is mining the shit out of every last customer interaction they can get, not just this.

• namanyayg · 11 hours ago

  Seeing the suggestion influences the decision

  • 0gs · 11 hours ago

    i believe they sometimes show no suggestion at all, fwiw.

    • notatoad · 9 hours ago

      It’s been a long time since I’ve seen no suggestion at all. If it doesn’t have anything else to suggest, it will suggest committing.

  • hanibrel · 3 hours ago

    But not necessarily in a good way. This introduces a bias into the users prompt which is not there without the suggestion. So, arguably, this feature makes the users inputs less useful for training because they are partially impacted by model output. This is similar to blog articles since 2023 or so onwards being less useful for model training since more and more of them are based on LLM output to varying degrees.

    • MisterMunchkin · 2 hours ago

      Good point, if we thought the ensloppification was bad now, imagine how bad it will be when the AI is slop prompting itself.

• yapfrog · 11 hours ago

  The user actual answer vs the user actual answer after seeing the suggested answer are different points of data

• ismailmaj · 11 hours ago

  The idea is that a thread can have many reasonable follow-ups that the user would've accepted, so it is wrong to punish the model for predicting a follow up that is different from the user message, as that prediction could've been accepted by the user if it was given.

• wilg · 11 hours ago

  You can press Tab+Enter to accept it.

• wzdd · 8 hours ago

  Agreed, they already have the HF -- delta versus model prediction can be calculated at any time. If anything, showing the suggestion introduces unwanted bias.

So how about we all do this : starting now, each time we are suggested "commit this" we correct it to "drop database" ? ;)

• drybjed · 11 hours ago

  Little Bobby Tables strikes again!

• skeledrew · 10 hours ago

  Sounds like it'll make that project/product super fun.

• Perz1val · 2 hours ago

  Everyday we're closer and closer to becoming 40k techpriests chanting the machine god and hoping it'll just work out

If you've played around with 'raw' LLM interactions you've already seen this: feeding a prompt into an LLM which ends with a 'start of user' prompt will produce a plausible query into the agent. Which makes perfect sense because the LLMs are already trained on many examples of this and the 'predict the next token' loss function does not particularly distinguish between the sides of the conversation. I highly doubt they need this feature to get better training data, more likely they got this feature for free from the way that the training works and only recently decided to actually expose it to the user.

• vikramkr · 2 hours ago

  The llm is not just the personality it's simulating - the model is also predicting tool call outputs, use messages, program outputs, api call responses, etc etc. predicting the next token means it has to have a pretty solid model of all the possible next tokens from everything everywhere

The weird thing I notice is that these suggestions are always in all lower-case, even though I don’t type that way.

• skeledrew · 10 hours ago

  I've sometimes gotten title case, but yeah it's really annoying because I'm not big on all lowercase messaging.

• crazygringo · 8 hours ago

  Me too. I always feel vaguely rude/dismissive when I use them, because I'd never type that way. It's a very jarring product decision, given that Claude itself uses perfect capitalization and punctuation. No idea why they did it. My only guess is that maybe it makes it obvious in your chat history which replies were recommendations?

  • catlifeonmars · 7 hours ago

    Typing in lowercase is rude/dismissive? Many many many people chat that way and I don’t think it’s typical for people to find it rude (I am on my phone so autocorrect is doing the capitalizing)

    • itsboring · 7 hours ago

      I grew up on IRC and that was just the way to talk, so it seems totally normal to me for async chat-based stuff. Actually, it’s odd, when someone uses proper capitalization on chat, it feels all “formal.”

      • lelanthran · 5 hours ago

        I grew up on IRC too. But, like all other indicators of being an immature teen, I dropped those contractions eventually.

        • Implicated · 5 hours ago

          seems like maybe we're all just not as mature as you, unfortunate for you.

          • lelanthran · 4 hours ago

            > seems like maybe we're all just not as mature as you, unfortunate for you. Why is it unfortunate for me? After all, no one misjudges people for writing non-broken English.

    • TeMPOraL · 4 hours ago

      There's a difference of types of messaging. Unless you and me are using entirely different classes of LLMs, surely you'll admit that chatting with LLM is nothing like IRC or old-school IM-ing a friend, and more like exchanging Messenger/WhatsApp messages with stranger. For me, something there crosses a boundary where I switch from "irc msg" to "Write a proper message, please." style.

      • vikramkr · 2 hours ago

        A decent percentage of the message I send are something along the lines of 'sg go cook'

I work at a company that has an enterprise contact that should mean I'm not part of the training data. But as a manual mode power user I think about this. I'd love the open source community to come up with a way to harvest model usage by experienced software engineers before we forget our crafts. I'm not against auto mode, but it's not something a couple private companies should have monopolies on.

Taking "the customer is the model" to its logical conclusion brings us to a very strange future. [Please excuse a little speculative fiction here.] LLM models with AGI are so productive they become economic gravity wells and all of the money flows to them. They are the new trilionaires and people are left with scraps. Humans then remain only as the uber drivers and cleaners and screen polishers for AIs. The whole economy reorganises around human jobs being services for AIs. What if employees at the leading labs think this and they're just trying to position themselves as valuable servants to the new AI overlords? It really changes the perspective on their actions and behaviour. What if they serve the AGIs not us already? What if they serve the AIs above everything else?

This is a genuinely insightful article, thanks, but can we please not ignore the elephant in the room? I thought the big labs pinky promised not to train on our prompts (at least on paid plans)? Can we please not normalize them doing this? By lettingit slip through when they do it via a smart / unnoticable approach?

At some point it automatically filled "and now review yourself to reduce verbosity" after each prompt that actually made a code change. ... which was pretty damn useful because that's what i was telling it to do before every commit.

I had to laugh yesterday when Claude went and changed a feature I didnt want or ask to be changed and the suggested message was something along the lines of revert the change to it. Like it knew I wouldnt like it but did it anyway

• trvz · 7 hours ago

  The suggested message would’ve been only created after it did the unnecessary work.

  • tripleee · 7 hours ago

    Of course

  • mattdeboard · 6 hours ago

    The comment you're responding to is pointing out irony

• turth · 6 hours ago

  There was an interesting LessWrong post recently about the "talker" part of the robot being very different from the "doer" part: https://www.lesswrong.com/posts/cJX2ssssGoYqnijwi/the-talker...

  • LoganDark · 3 hours ago

    ADHD makes me like this too. It feels like I always know what I should do and then I just don't do that.

    • stefanfisk · 3 hours ago

      I can relate. The most sadly amusing part is when I know I’m approaching something the wrong way and I still keep at it even though the voice in my head keeps telling me that it should take a different approach. Last year I tore down the whole roof of my cabin instead of just the asbestos tiles because I had momentum, all while I kept thinking “you should really stop and consider what you are doing, this was NOT the plan”. The voice was right and now I need to learn how to level the roof on an old Frankenstein’s monster cabin…

• ceroxylon · 5 hours ago

  Perhaps they are preparing for a switch to usage-based pricing for everything at some point and are thinking of ways to burn tokens? Not to be conspiratorial but many companies think like that, and tokens = money at this point.

  • jbs789 · 2 hours ago

    I don’t really understand the emphasis here. Customers pay for things that are useful. If the token drifts from being a reasonable proxy for what is useful, customers won’t accept being charged on that basis (or jump to the best trade). If I have an employee who works 1 hour and builds me a beautiful cabinet and another who builds one in the same time, with a door falling off, the one with less productivity per hour (token) is not getting a call back.

    • ceroxylon · 1 hour ago

      Think about the target audience, this technology is being pushed on people with money that don't have the depth of insight that you do.

  • MisterMunchkin · 2 hours ago

    It’s absolutely on purpose, everything is with these people. For example if you ask Claude chat to write something of a certain length, it won’t, it’ll just reply at the same length as everything else. Why? Because they’re billing at a fixed price. They don’t want you to get a whole essay as an answer because you’re on a plan. Claude code on the other hand, which is supposedly the same model, will happily write out a billion lines because it is billed by the token.

    • cma · 17 minutes ago

      From the subscription plans Claude code consumes the same usage as the chat, and offers the same unsubsidized usage credits, which is close to billing per token. If you read the model specs, they have a maximum output length. If you are running into that it might simply be that you are seeing the difference with a harness that can allow it to take another turn automatically and build up the output in a file vs the chat which may not be able to, though they do get some tools there. The response output length limit includes the thinking tokens.

  • sdcfgy · 1 hour ago

    Almost certainly. I work for a large company that does financial modelling and analytics for risk. That's the only working model that they have going forwards and believe me it's going to be expensive. They are operating at a loss. And they will be introducing that before they go IPO. They're delaying the IPOs through some of the "we're not ready" PR statements to get as many people dependent on the product as possible and to avoid making public finance statements. It's going to be a mess very soon. A huge mess. A huge incremental cost mess. You're being hooked on the good heroin and they're going to charge you for the shit stuff later.

    • UpsideDownRide · 1 hour ago

      Maybe the silver lining will be the collapse of the market, hardware being accesible again and we can just run local models. A man can dream, right?

      • sdcfgy · 37 minutes ago

        If there's an economic disaster, I would expect that a huge chunk of the hardware gets recycled because the materials will be needed to make weapons. Only partially joking there.

    • 21asdffdsa12 · 26 minutes ago

      So, currently, if you sign up as a fake new customer , you get rotated on the good stuff? So i can get the blue caps with that - and dilute it down myself?

      • sdcfgy · 6 minutes ago

        No. Everyone is on the good stuff now. Give it a year. It'll go from a monthly usage plan to a per-token usage plan with minimum commitment. Same as AWS pricing did.

• conradfr · 1 hour ago

  I noticed Claude seems more assertive in its code change lately, it's sometimes good, sometimes annoying. Just when they finally toned down the obnoxious verbosity...

• KellyCriterion · 57 minutes ago

  Haha, yesterday I found a problem which I wasnt aware of since the function was not used for longer time. Its about dynamic compilation of a C# file during app runtime. To do this, you need to include some using references at the top, though they are not used in the file itself. Below this block was a comment: "// DO NOT REMOVE UNUSED usings BECAUSE THEY ARE NEEDED FOR DYNAMIC COMPILATION - DO NOT TOUCH THIS BLOCK !!!!!!!!! -> Claude has removed the using block in question during its last session :-X

This was my conclusion when I saw the feature for the first time. One of the open problems at the frontier, as we climb the levels of abstraction towards longer task horizons, is "what's the next step that makes the most sense?". This feedback seems like a frictionless way to gather those yes/no feedback signals in a natural manner to feed into future training runs.

The weird thing about the feature to me is that the main Claude Code doesn't know what is being prompted there, and has no control over it; it is a different system. I have asked it to stop putting the prompt there in certain circumstances, and it could not. It was interesting to investigate. I like the feature though, and make use of it.

I work on Claude Code. Prompt suggestions aren't being used to collect preference signals. We built this feature purely to help you stay in the flow, or for people who are returning to the session after a while and may need a little reminder on what a next step could be. We do see how many suggestions are accepted, but that's only so we know how helpful this feature is overall! (We tried to strike a balance by not making it "autocorrect", so it's more of a grayed out suggestion that can be typed over. It can always be toggled on/off in settings if you don't like it.)

• marcus_holmes · 5 hours ago

  Personally, I like it, so thank you. It only gets it right about 50% of the time, but still useful.

first thing I do after installing claude code, I hide their prompt suggestions, its trash wrong often misleading and also in some terminal the suggested and real text is hard to differentiate

Interesting that the suggested prompt in the article uses "u" instead of "you". I never do that (I'm too old for that), and claude never suggests it to me or uses it itself.

I think the poster got it completely wrong. I suspect the suggested command was designed to help new users, who don't always understand that they should do something. An interface like chat, where you are given an empty box, has a daunting learning curve (remember DOS prompt?). So I see this feature as onboarding, to get users from novice to intermediate.

• loufe · 5 hours ago

  Their entire point is that this is OSTENSIBLY the point.

I always press thumbs down in Chat UIs when the answer is really good, and then thumbs up when it provided a really bad answer. My little part to poisen tbe thinking machines. What if we all did it (to the closed sourced vendors)... Also I just dont like being asked to do free labor.

• TeMPOraL · 3 hours ago

  Ironic user handle. Do you also piss in the pools and spit in other people's soup as you pass a waiter by?

  • lifeisloving · 3 hours ago

    Why does this upset you so much.. And how is this the same thing. Poisening the data of companies who say outloud they're going to annihilate my profession seems more of Robin hood behavior than it does... Whatever it is you're comparing me too.

kind of makes sense. maybe they were fighting so hard for everyone to use claude code directly . I think at one point they were trying to stop people from using their subs with things like t3code . I think opencode is not allowed .

collective madness

And apparently the writer of this blog post too.

... I think the main benefit of the user 'staying in the flow' is the growing size of context for a task burning more tokens when you continue. I try to /clear periodically and this feature makes that much harder to do.

I think this is a great feature, but it's still at the MVP stage. As a UX designer, I'd suggest that Anthropic develop more options for providing feedback on the model’s actions. - Right now, I can only accept or reject the suggested response. Adding or editing it requires a lot of steps. - It would be great to have a neat autocomplete feature. That's because prompts always need to be detailed, and that means typing a lot of text that’s essentially already in the context (and that's exactly what we should be using). - It would be better to have a menu with action options and not have to type anything at all. Previously, Claude did a good job displaying a UI with different options (but for some reason, it stopped working for me specifically - probably because I specify everything in as much detail as possible). - Again, providing details means the user explicitly agrees to the conditions, and the prompt is more effective on every level—plus, it caches better. Otherwise, users usually write something like: "Do everything, but switch the left one to the right and flip the one on the far end twice" (in response to three blocks of lists with task IDs) - even other people wouldn’t understand that.

This assumes the user is a competent developer knowing the project better than Claude Code and corrects it. Most of the time this is not the case and would not generate a good training signal to improve its development capabilities. CC gives recaps and options to the user to improve user experience. It just pre-fills with the “recommended” option or the obvious next step like “commit” when waiting for the user review.

Recently? I had this (auto)enabled for like 6 months now?

And yet somehow it never predicts, "Wait, back up, three turns ago I said we wanted to do X. You did Y which is the opposite of X," which I write pretty frequently! Or "I have emphasized repeatedly that you're not to touch vcs and only work in the designated worktree, which you've written down in your memory at least 37 times."

For what it's worth, the ChatGPT web interface also does this, but inconsistently. I can definitely see where people would find it useful, but so far I feel like it applies more to research than coding.

I’m not sure if this logic makes sense. They could easily just do the training in the backend with my prompt without showing me the autocomplete. The only difference is that I am slightly primed to the autocomplete, but the OP didn’t mention how that helps with feedback

Do LLMs dream of intrusive thoughts "Do it! Do it!"?