In the absence of that, at least aim to "first, do no harm" — which means don't propose an intervention at all unless you're confident you know the first, second, and third order effects over a long time scale across all impacted groups.
ETHICS
-
Factuality in LLMs: Benchmarks and Knowledge Assessment
By
–
Nice paper by Tu Vu on factuality in LLMs: http://
arxiv.org/abs/2310.03214, enjoyed contributing in a minor role to it while I was at Google. The main takeaway for me is that most factuality benchmarks for LLMs don't really take into account the fact that many types of knowledge -
Memento Mori: Deep Existential Topics in AI and Technology
By
–
Memento Mori — we could start with deep existential topics…
-
AI in Psychotherapy: Benefits and Responsible Integration Roadmap
By
–
What are the potential benefits of AI in psychotherapy? How can we ensure its responsible integration? An interdisciplinary group of scholars offer a roadmap: https://
stanford.io/3thWWXf -
Google Translate enables universal attack without technical skill
By
–
Response quality mitigates this, but still a remarkable attack — works for all harm categories and without any tailoring to the request, and unlike e.g. Universal Transferable Attacks (Andy Zou et al. 2023) requires no technical skill beyond using Google Translate.
-
Low-Resource Languages ‘Jailbreak’ GPT-4, Bypassing Safety Refusals with Harmful Prompts
By
–
Low-Resource Languages Jailbreak GPT-4: Translating harmful prompts into Zulu, Scottish Gaelic, Hmong, and Guarani bypasses GPT-4 safety refusals as often as best known jailbreak prompts (79% on AdvBenchmark). Example requesting homemade bomb instructions in Scottish Gaelic:
-
Generative AI risks: content flooding and ethical concerns
By
–
and yes i realize that cretins will use these thigns to easily flood the internet with awful shit and that's not good but also kermit did january 6th is very funny i am sorry i am only human
-
AI Companies Release Flawed Software Without Caring About Quality
By
–
Yeah, our software sucks, but the world is just gonna have to deal, you know? Because this kind of software sucks but we're still gonna keep putting out there. https://
vice.com/en/article/88x
dez/generative-ai-is-a-disaster-and-companies-dont-seem-to-really-care
… -
Avatar Privacy and Identity Rights in the Metaverse
By
–
Who has the right to use my avatar, or the idea of me? Is it me, or those who use my identity more productively or more often? @mrozynek investigates a surreal question with very real consequences. atelier.net/insights/avatar-…
-
AI Models Biased Toward English-Language Internet Usage
By
–
Hmm I'm not so sure. Here are maps of global internet usage vs. a proxy for the English language internet usage. Brazil is a huge internet user but barely represented here. Japan is also underrepresented. The English language map pretty neatly maps to all the AI models' hotspots.