I am happy to announce our workshop in #DIS2024 "Towards Robot and Technologies that Touch Well", the first workshop in DIS for over a decade that addresses touch. You can read more about the workshop and how to register here: https://t.co/gJ9LlHDxyZ See you!
For #ttrpg#gamedesign purpose: what icon would you use for "effort" and "favour"? Both are game resources, effort improves skill checks, favour is meta currency. Theme is middle-eastern & india inspired fantasy.
Citation alert from Google Scholar for the first citation on a paper as first author: A run-away entry in the bibliography, not actually mentioned in the text :'(
Please don’t hesitate to get in touch if you want to apply for this to work in a feminist design research group working on complex topics, in from the margins!
Hebrew speaking researchers, can you recommend any *offline* transcription tool that supports Hebrew? I tried "Whisper Transcriptions" from Apple Store but it's editing interface is terrible...
Hebrew speaking researchers, can you recommend any *offline* transcription tool that supports Hebrew? I tried "Whisper Transcriptions" from Apple Store but it's editing interface is terrible...
There was one low-hanging improvement that could have been done by publishing the zoom call link for each room online. All speakers were asked to join the call anyway for auto-transcription, the link might have well been available online.
Been reading a few #chi2023 folk tweeting disapproval with the online experience. Clearly it was minimal-ish but it’s easy to underestimate the insane amount of work (and cost) needed to do hybrid well. Personally, I don’t think the technology is there yet.
Can one create a "safety net" around LLMs (the likes of Jurassic-2 or GPT-4) to make sure that LLMs are "aligned" with human values? there has been a lot of chatter around "jailbreaking" ChatGPT and the question is whether through further training one could "tame" the machine?
New paper! Does alignment of LLMS trained on massive corpuses that inevitably contain undesirable behavior is feasible? We proved that the answer is not 🛑,
and that the popular RLHF alignment technique can make the situation worse! More details in https://t.co/i0f2Kngc3C
don't be confused, I think it is a great idea. Although sometimes I would like to imagine that my program works even if it is leaking some memory here and there.