Show HN: How Stale Is Your AI? Release age and training cutoff for 20 models

gjskngnf•about 1 hour ago

I remember when the US captured Venezuelan president Maduro, and when I posed a prompt related to this, the model said that’s pure fiction. I told it to double check. Still didn’t want to entertain the idea. It only acquiesced when I specifically directed it to check Reuters. I haven’t noticed this problem in months. Model cutoff seems to be less of a problem these days.

super256•about 1 hour ago

It's a "problem" of compute, I think. If you query without an account on ChatGPT you will see the model look up less stuff and research less, than when you have a paid account and choose "medium" or "high" in the effort slider.

Which makes sense, because of you have looked into search and crawlers you notice that search is actual quite expensive (which is why e.g. Kagi charges a few bucks for search every month).

InsideOutSanta•about 1 hour ago

Came here to say the same thing. Models used to rely heavily on world knowledge from their training data. They are now much better at tool use and deciding when to research a topic, rather than just answering from memory.

I wonder how much that extends to using LLMs for programming. I assume most knowledge of programming language syntax still comes from training data.

jasonjmcghee•about 1 hour ago

It still matters, but in the age of good reasoning, tool use, and web search, this is much less of a problem than it used to be.

dominotw•about 1 hour ago

all the reasoning still comes from pretraining data

binlog•about 1 hour ago

Says who? Models can also use results from tool calls in their reasoning loops.

dominotw•41 minutes ago

some logic here

https://www.lesswrong.com/posts/HYCGA2p4bBG68Yufh/thinking-a...

bigmadshoe•about 1 hour ago

I think they mean reasoning their way to the need for a web search.

jasonjmcghee•17 minutes ago

Or other kind of search / knowledge acquisition / computer use etc to get the information needed

ryanschaefer•about 2 hours ago

Do people prefer the new flat style LLMs are producing? I don’t mind it as much as the gradient theme they were pumping out previously.

VCFundedGenYer•about 1 hour ago

I remember running the docker container for ollama and its knowledge cutoff is somewhere in 2023 still. That's unacceptable.

jasonjmcghee•9 minutes ago

ollama is just an inference engine - it just runs models.

it must ship with some default old model if you didn't need to explicitly download one

speedping•about 1 hour ago

Pre-AI internet data is like pre-war steel

The slop would multiply if we keep feeding it to new models in a loop

j45•about 1 hour ago

Depending on the use case certain models very well remain as or more reliable for certain tasks.

delichon•about 1 hour ago

After Trump's last inauguration, ChatGPT would still tell me that Biden was President of the US. I understand that the training cutoff was before Biden dropped out. But it knew, or should have known, the current date and that there had been an election since its last update, but it didn't qualify the answer. When I asked it to search the web, it got it right. The moral I took away was to always ask for the search whenever I ask about current events. I do that so routinely that I wouldn't know if this problem has been fixed. I suppose that failing to update my priors per individual model release is a form of bigotry against a widely hated class.

andai•about 1 hour ago

ChatGPT recently started web searching for for basically every general knowledge question, which I found quite odd. Maybe an overcorrection to the issue you were having?

Show HN: How Stale Is Your AI? Release age and training cutoff for 20 models

⚡ Community Insights

Discussion (16 Comments)Read Original on HackerNews