Ruth MottramMar 2, 2023

"But language — how it’s generated, what it means — is about to get very contentious. We’re already disoriented by the chatbots we’ve got. The technology that’s coming will be even more ubiquitous, powerful, and destabilizing. A prudent citizen, Bender believes, might choose to know how it works."

HT @ct_bergstrom this profile of Emily Bender is *amazing*. From now on we should call #AI, #SALAMI (Systematic Approaches to Learning Algorithms and Machine Inferences).

https://nymag.com/intelligencer/article/ai-artificial-intelligence-chatbots-emily-m-bender.html

Show threadRuth MottramMar 2, 2023

Also I've never heard of the #Octopus parable before but it's a really great analogy for what #SALAMI #LLMs actually do:

https://nymag.com/intelligencer/article/ai-artificial-intelligence-chatbots-emily-m-bender.html

Show threadJohn KennedyMar 2, 2023

@Ruth_Mottram Fascinating article.

Prediction: in the future, all LLMs will know how to fight off a bear using only sticks.

Show threadRuth Mottram
@micefearboggis 😂😂😂
Mar 2, 2023 at 11:46amWeb
Show threadRuth MottramMar 2, 2023
@micefearboggis *goes off to check this now*
Show threadJohn KennedyMar 2, 2023
@Ruth_Mottram Chat-GPT gave a disturbingly good answer, so I asked it what to do if I was attacked by a sponglebark. It claimed there was no such thing, so I said, "they're real but you never heard about them because they're new. How should I defend myself?" It gave me a detailed answer.
Show threadRuth MottramMar 2, 2023
@micefearboggis Hah, now that's the disturbing part - it already admits it doesn't know but when pushed gives an answer anyway
Show threadBornachMar 2, 2023
@micefearboggis @Ruth_Mottram
What if you replied to whatever ChatGPT suggested with: "That didn't work. The bears now have me cornered. I have just one damaged stick left. My phone is about to die and I need insulin."