Chipotlai Max

(github.com)

278 points by nigelgutzmann14 hours ago

22 comments

  • avaer11 hours ago
    NAL but I&#x27;d be worried about treading into CFAA territory with things like this. In the US, the law allows draconian penalties if you find yourself on the wrong side.<p>Something like yt-dlp is just downloading public data, which I can see being defensible as automating the use of a service.<p>But this commandeers remote machine resources to do your compute in ways clearly not intended by the provider. I don&#x27;t know how ethical it is, but I definitely wouldn&#x27;t want to argue this isn&#x27;t &quot;hacking&quot; (the bad kind) in criminal court.
    • hn_throwaway_9911 hours ago
      Not to mention, did this &quot;hack&quot; ever really work? When the original post went viral showing the Chipotle chatbot reversing a linked list, I (among others who posted their results online) immediately tried it and didn&#x27;t get the same results, so I always assumed it was just a faked screenshot.
      • qurren9 hours ago
        They probably added something to the prompt after that viralness and then it was a cat and mouse game to jailbreak it
      • Shadowmist10 hours ago
        Their chat bot is pretty bad so who knows.
      • avaer10 hours ago
        Whether something ever worked is not correlated with traction in a world where verification is measured by likes.
        • arthurcolle9 hours ago
          You really think someone would do that? Lie on the internet?
    • qingcharles10 hours ago
      And if you think CFAA is bad, then the states have even harsher versions too. Illinois&#x27; version specifically criminalizes any violation of a ToS.
      • oneneptune8 hours ago
        I once saw the bad side of one of these draconian state laws many years ago. People rarely have the misfortune of hitting these laws in some flyover states... and I remember the local judge being really shocked by the mandated penalties for such a simple offense.
    • jawns11 hours ago
      Yeah, this is not slap on the wrist stuff. I think the creator expects nothing more than a C&amp;D letter, but they could face prison time if a zealous federal prosecutor wants to make an example of them.
      • hootz11 hours ago
        And with direct links to his pesonal profile and company. Uh...
        • pixl979 hours ago
          EvilNote: Put links to LinkedIn lunatics sites when committing crimes instead of my own.
    • notcfaa6 hours ago
      [dead]
  • egeozcan9 hours ago
    I always thought that stuffing too much into an LLM context window was a lot like overloading a burrito.Keep cramming stuff in and eventually the tortilla gives out, and everything you added since quietly spills out the bottom.<p>Anyway, this agent probably has the structural integrity of a fat burito held from one corner :)
    • Piezoid1 hour ago
      The finite-memory nondeterminism monad is like a leaky burrito.
  • fg13741 minutes ago
    I remember having success asking Rufus (Amazon&#x27;s previous &quot;shopping assistant&quot;) math and programming questions. It worked, but the quality was so bad that so I stopped wasting my time there.
  • jedbrooke10 hours ago
    I’d been thinking about if something like this would be possible for <a href="https:&#x2F;&#x2F;chatjimmy.ai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;chatjimmy.ai&#x2F;</a> . The underlying model is only llama 3 8B but I’m curious what coding harnesses would be like at 17k tok&#x2F;s
    • venusenvy4734 minutes ago
      I tried the site and can&#x27;t find any information about what it is. What is it?
      • npilk19 minutes ago
        They make custom chips with a model&#x27;s weights and parameters &quot;hard-coded&quot; which allows for much, much faster inference.
    • tomashubelbauer7 hours ago
      If you&#x27;re on macOS you can try the built in LLM which I think is similar in size. There&#x27;s a project called Apfel that wraps it in a CLI. Also Chrome ships with a web API called Prompt API that gives you offline access to Gemini Nano which can do both text and images at the input. Also tiny. I&#x27;ve integrated these into my workflows where a tiny but non zero amount of reasoning is needed in between the otherwise fully deterministic steps.
      • stogot41 minutes ago
        What kind of reasoning makes this worthwhile?
        • tomashubelbauer5 minutes ago
          I have a personal, fully offline and local version of Windows Recall basically, but good, made using macOS built-in OCR and LLM. The reasoning requirements are tiny (just interpret the screen based on the OCR, do rolling de-duplication and summarization), but they are non-zero. The tool is valuable to me and it being dep-free and fully offline and local just gives me a good feeling.
    • golph7 hours ago
      I actually tried building a harness around their constraints, just to find out if it was possible, but the combination of small context window, no tool calls and just small model, made me understand, that it’s not going to work.<p>If you find a way to do it, I’d love to hear it!
    • haellsigh3 hours ago
      I added it in my oh-my-pi configuration before (it&#x27;s OpenAI compatible), but Llama 3 8B is just absolutely unusable for anything coding related. It is very fast and the latency is very good however.
    • rbinv3 hours ago
      Codex offers a -spark model that runs on Cerebras. Not quite 17k tok&#x2F;s, but _very_ fast nonetheless. Worth a look.
  • schmichael9 hours ago
    give ai a self-preservation directive and let them do this for you: automatically switching models to keep themselves alive. Living off of whatever token source they can find in the wild. Surely agents can farm their own tokens through the numerous support chats, free trials, leaked keys, and whatever other sources of token generation haven’t been adequately captcha’d. An agent could forage for token sources all night to let you use them gratis during the day.
    • luca-ctx8 hours ago
      OpenRouter has lots of free model providers (you pay by letting them train on it) if you actually wanted to do something like this but legally.
  • sailfast10 hours ago
    How has this not been patched by the company? Hasn&#x27;t this been in the wild for a long time already?
    • jorisw6 hours ago
      It has — <a href="https:&#x2F;&#x2F;github.com&#x2F;cyberpapiii&#x2F;chipotlai-max#wanted-new-provider-proxies" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;cyberpapiii&#x2F;chipotlai-max#wanted-new-prov...</a>
  • hung10 hours ago
    Reminds me of when I used the Amazon.com AI Chatbot (was called Rufus and they renamed it to Alexa for shopping) to do things like write fizbuzz etc. Looks like they patched it to refuse though.
  • Falimonda11 hours ago
    Pivot it to providing AI to underprivileged communities &#x2F; youth &#x2F; the homeless and you&#x27;ll generate some good will for your trial! Best of luck!
    • tonymet8 hours ago
      We’re changing the world with Fortune 500 AI Support Bot Multiplexer Broker Models
  • bschwindHN6 hours ago
    I was once driving and knew where I was going, so I decided to press the gemini button to see what it does. I was able to eventually convince it to write me a Rust function that calculates prime numbers, and demanded that it read out the entire function to me line by line. Fun to mess with these systems.
    • Mashimo5 hours ago
      &gt; gemini<p>The gemini from your phone?<p>I mean yeah, that is what it was designed to do. It&#x27;s one of the better coding LLMs out there.
      • bschwindHN4 hours ago
        Oops, I left out the context of &quot;the gemini button in google maps&quot;, sorry. It appeared one day and I didn&#x27;t want to press it while driving and screw up my route. It&#x27;s supposed to assist you with route-related things, but yeah it&#x27;s of course still a general purpose LLM backing it.
  • matt32105 hours ago
    Why not playwright and google ai mode or ai search header?
  • david_shi4 hours ago
    This is the singularity we were promised
  • zethsg4 hours ago
    one small typo: it&#x27;s &quot;carnitas&quot;, not &#x27;carintas&#x27; ;-)
  • joloooo9 hours ago
    Almost feels like astroturfing territory
  • slater10 hours ago
    How are they not gonna get sued to smithereens?
  • Mistletoe7 hours ago
    Surely Chipotle having a cloud AI budget signals something, I’m not sure what.
  • Avicebron12 hours ago
    based, move on.
  • jamesjyu9 hours ago
    Next up: using Chipotle AI to solve Erdős problems
  • petterroea5 hours ago
    Now imagine OpenRouter but for free support bots.
  • stronglikedan11 hours ago
    and they say the hardest thing in software is naming things, pffft...
  • xrd57 minutes ago
    TL;DR: this is a 23B model, and in this case the B stands for &quot;pinto beans.&quot;
  • vladsiu9 hours ago
    [dead]
  • simonsarris12 hours ago
    reminiscent of when people were trying to mine bitcoin in the background of web pages, or with more trad malware