#halucinations

Larvitz :fedora: :redhat:

@larvitz.burningboard.net.ap.brid.gy

9 months ago

Tried asking a Large Lying Model (LLM) something about firewall configuration on Ubiquity network devices in combination with WireGuard VPN.

I was very specific about which device it was, which software version it was running and what I wanted to achieve. The models answered confidently, but […]

Adam Shostack :donor: :rebelverified:

@adamshostack.infosec.exchange.ap.brid.gy

9 months ago

Database of 117 instances of lawyers using #LLMs inappropriately https://www.damiencharlotin.com/hallucinations/ #halucinations #ai

Paula Salamone

@paulasalamone.bsky.social

1 year ago

@linkopinguni.bsky.social @bodyrepnetwork.bsky.social #schizophrenia #halucinations #science #fMRI #EEG #spinalcord #bodyrepresentation #neuroscience

Michael Washington

@adefwebserver.com

1 year ago

Do you think the models *ever* will be 'smart enough to be stable'? I think #halucinations are a feature not a bug. It's like being annoyed at a hammer because it's heavy. To me the LLMs simply perform a 'calculation'. "Truth" is not something that can be calculated in my opinion.

P Maroney

@patrickmaroney.bsky.social

1 year ago

Google's AI tells users to add glue to their pizza, eat rocks and make chlorine gas Social media has been flooded with bizarre and dangerous advice that appears to have been made by Google's new AI feature.

Google's #AI tells users to add glue to their pizza, eat rocks and make chlorine gas | source Live Science
➡️
www.livescience.com/technology/a...

#hightechheadlines #ai #google #halucinations