A big problem with generative AI is evaluating the information. You can't use the CRAAP test bc generative AI is predicting text after being trained with *many* sources. Is the information current? Is the author credible? Is there a conflict of interest? You can't answer these questions. - ThreadSky

heydebigale.bsky.social • 50 days ago

A big problem with generative AI is evaluating the information.

You can't use the CRAAP test bc generative AI is predicting text after being trained with *many* sources.

Is the information current? Is the author credible? Is there a conflict of interest?

You can't answer these questions.

Comments

kjhaxton.com•50 days ago

I’m encouraging a ‘fact checking’ critique approach, can you verify each bit from a source that would pass the CRAAP test? Then looking into blackout poetry approaches to remove all of the fluff language that pads out the responses to prompts

heydebigale.bsky.social•50 days ago

(But maybe that’s the point? To show students how this is not the way?)

kjhaxton.com•50 days ago

Yes, if it needs to be used then prompting for structure or suggested topics then doing the research works better. The bit that annoys me is the amount of time spent verifying references in submitted work actually exist.

heydebigale.bsky.social•50 days ago

Oh my god it’s just so time consuming and why use genAI at all when you could just do research the old fashioned way

racicotleanne.bsky.social•50 days ago

I was literally talking about this with my new coop student and a TA. I'm team "just no AI" exactly because we can't evaluate the reliability of the information and how the LLM are built (sometimes trained by low wage workers with no labor protection). But it feels like a hot take?

racicotleanne.bsky.social•50 days ago

And the thing that's hard is technology is evolving much too fast to adapt our guides to help students navigate resources and protecting integrity/our IP within all this. And EdTech is fully pushing AI-everything, my institution is encouraging us to develop ways for us to use it for teaching *sigh*

heydebigale.bsky.social•50 days ago

Nope nope nope

heydebigale.bsky.social•50 days ago

This issue is talked about a bit in the book AI SNAKE OIL but I want to know more!

thatlibraryman.bsky.social•50 days ago

I've been quietly working on a new approach to evaluating information for these very same reasons, encouraging students to not only assess the content but understand how the information has arrived at them, not just in terms of GenAI but also algorithms that are involved in other discovery methods

heydebigale.bsky.social•50 days ago

I’d love to hear more about this! I’m very against genAI for a multitude of reasons (copyright violations, climate issues, etc) but I’d like to discuss the information literacy aspect with my students in an informed way.

thatlibraryman.bsky.social•50 days ago

I completely agree with you on the issues with GenAI. Unfortunately though our students are using it anyway, so it's important we arm them with the critical skills to identify its shortcomings themselves.

Drop me DM if you like and I'd be happy to chat further and share my ideas 😊

ijayas.bsky.social•50 days ago

theokane.bsky.social•50 days ago

It's almost like the people creating AIs have forgotten that they are built by humans and trained on stuff created by humans and that human-created computers aren't infallible.

heydebigale.bsky.social•50 days ago

I mean they must know!

But it is super frustrating nonetheless!

Comments

Posting Rules

Reply