What will be the most common word we use for processing text with large language models?
Basic
29
Ṁ724
Jan 1
47%
ask ChatGPT
35%
ask an AI
9%
Ask Chat
3%
GPT it
3%
AI that
2%
Other

Looking for the equivalent of "google it". Example usages:

"When I need something explained to me now, I don't google it, I [verb] it"

"The application takes the user input and then [verb]s it to get a summary"

"The startup is basically just [verb]ing it under the hood with clever prompt engineering".

If we don't end up with a single word for this, then I will resolve the market as unresolved/unknown.

Will try use my best judgement to resolve — if we don't have a clear contender or winner within a year, I will extend the market deadline.

Get
Ṁ1,000
and
S3.00
Sort by:

@Mag Can you please resolve this market? I appreciate it very much!

@Nps ngl, at first I thought this was meant as asking a twtich/youtube's stream chat.

@NoyaV I would like to see an LLM that responds by constantly spamming thousands of only tangentially related messages to your prompt like you would see in a popular streamer's twitch chat.

I have seen multiple uses of "Ask ChatGPT" and "Ask an LLM" in the wild on twitter now, I hope someone comes up with something better, feels so clunky

i'd propose that a compatible and optimal resolution system would be to take the distribution from an up to date language model at the time of resolution.

@L Note that this requires there to be an up-to-date lifelong learning or whatever language model that has high quality probability distributions for text from the appropriate context at the time. speaking of which, what contexts do you care about in terms of where the words are getting used by humans?

@L based on the lag between the training data and the most up to date available LLMs, I think this will be too long to wait — currently ChatGPT is trained on data up to mid-2021. Assuming whatever word wins out starts become used popularly around now to mid-2023, we shouldn't expect LLMs that have been trained on data containing this word until late 2024

@MagnusHambleton maybe, but if someone feels like making a market about whether training will become more real time, I sure would bet it would, multiple labs are talking loudly about lifelong learning right now.

@L I care about it in terms of humans explaining something to other humans — e.g. as in the examples in the market description. E.g. "What does your startup do?" "We make a customer service chatbot that takes customer queries, [verb]s them and provides answers, finetuned on previous customer conversations"

@MagnusHambleton Does this include "ChatGPT"?

@EliasSchmied Yes, I would like the word to be something usable more broadly, also beyond the GPT models. E.g. a startup describing what their product does under the hood would use this word irrespective of the specific language model under the hood (they might say "and then we just GPT it to get the summary" even if they are actually using PaLM)

Lelm = play on LLM acronym, Lext and Langomize are suggestions from ChatGPT. @LarsDoucet suggested GPT as the Kleenex version, but I am yet to hear it be used as a verb.

© Manifold Markets, Inc.Terms + Mana-only TermsPrivacyRules