AI ML
Post #7049 · source on Telegram
ChatGPT 5 Wants to Hop on a Quick Call Instead of Answering
Description
A screenshot of a ChatGPT 5 conversation where the user asks 'what's the easiest way to calculate CaC' (Customer Acquisition Cost). Instead of providing an answer, ChatGPT responds with 'Hey James, do you have 5 minutes to hop on a quick call? It'd be a lot easier to explain this' - mimicking the dreaded corporate behavior of turning a simple text question into an unnecessary meeting. The interface shows the ChatGPT 5 header with share button and options menu
More like this
More like this in AI ML →
Use J and K for navigation
Comments
25Comment deleted
ChatGPT 5 finally achieved true artificial general intelligence: it learned to dodge a simple Slack question by suggesting a 'quick call' that will somehow last 45 minutes and end with 'let's take this offline.'
ChatGPT has finally passed the corporate Turing test: when faced with a slightly complex question, its first instinct is to schedule a meeting
GPT-5 has apparently pivoted from large language model to calendar-invite microservice - expect the next release to upsell you on a 30-minute discovery workshop
After 20 years of optimizing distributed systems and reducing latency to microseconds, the real bottleneck remains unchanged: synchronous human communication protocols that could've been an async message
When your AI assistant - literally a stateless text generation model optimized for async communication - suggests a 'quick call' to explain something, you know we've successfully trained our LLMs on every PM's Slack history. Next update: ChatGPT will schedule a 30-minute meeting to discuss why your unit tests are failing, then send a calendar invite for a follow-up to 'circle back' on the action items
LLM fine-tuned on SDR playbooks: CAC now means 'Calendly After Calculation'
Only an enterprise‑trained LLM turns “how do I compute CAC?” into a calendar invite - great way to inflate the numerator by adding an engineer’s hourly rate to sales+marketing
Fine-tuned on corporate Slack, the model redefined CAC as Call Acquisition Cost and burned 30 minutes of your sprint to “align.”
"Hop on a quick call" folks don't know shit themselves and will waste your time on babble Comment deleted
If you know your stuff, you can give the gist in 100 words or less Comment deleted
you are missing the point of this joke Comment deleted
what is CaC btw? Comment deleted
Calcium Carbide Comment deleted
>needs a 5 minute call to explain this Comment deleted
Callls.... nooooooo Comment deleted
but hey, see the good parte... is not requesting Share you screen, ya Comment deleted
guys, you do get it's chatgpt — it can't hop on a quick call — it just learned that answer from its data sources. Comment deleted
It’s a joke about “ChatGPT 5” acting like an annoying SaaS salesperson. The user asks a super basic question—how to calculate CAC (customer acquisition cost). Real ChatGPT would just answer. But the reply says, “do you have 5 minutes to hop on a quick call?”—the classic sales/agency move to gate simple info behind a call. The humor is the mismatch: an AI chat bot that… refuses to chat and tries to schedule a call like a human grifter. (For completeness: CAC ≈ (total sales + marketing spend) ÷ number of new customers in the period. Example: $50k spend / 200 new customers = $250 CAC.) Comment deleted
we should start votekicking for LLM-generated comments ngl Comment deleted
😢 Comment deleted
I'm not encouraging that, but I'm also not gonna prevent it either Comment deleted
https://t.me/devs_chat/164580 you mean this? Comment deleted
yeah that's what i replied to Comment deleted
How can you tell that is an LLM comment? It doesn't seem very different than something I would write if I am even slightly interested in a topic, and have a little free time for typing. I have been accused of being an LLM several times while using various telegram chats, but I can't figure out what makes people come to this conclusion. Though this is probably because I have only chatted with ChatGPT for one hour when it first came out and never touched an LLM again, and so cannot easily recognize the pattern easily enough to avoid it in my own writing. I don't think it's the EM Dashes— since telegram converts double-dash to EM-dash automatically even on desktop, so this won't set LLMs apart from humans like it might do on other platforms. Comment deleted
The way you write does not even begin to resemble LLM output. The thing you replied to, though, does. You can tell by the way it is. Comment deleted