all your tokens are belong to us
SAFETY
-
Robot taxi companies use remote assistance despite autonomy claims
By
–
"By creating the illusion of complete autonomy, companies can fuel interest in their technology and raise the billions of dollars they need to build a viable robot taxi service." Waymo and Cruise have recently admitted they use remote assistance. Zoox is more up front about it.
-
Rigorous Pre-Release Testing and Safety Evaluations Framework
By
–
We are also extremely proud of how rigorous and thorough our pre-release testing process is, including a full stack of frontier risk evaluations according to our Preparedness Framework and external red teaming. Read more safety work and evaluations in:
-
OpenAI Releases o1: Advanced Reasoning Model with Improved Safety
By
–
Finally o1 is out – our first model with general reasoning capabilities. Not only it achieves impressive results on hard, scientific tasks, but also it gets significantly improved on safety and robustness. https://
openai.com/index/learning
-to-reason-with-llms/
… We found reasoning in context about safety -
AI risks: misinformation, manipulation, and harm concerns
By
–
I hear you but I'm worried it will increase the risk of misunderstanding/misinformation/harm/manipulation
-
AI Metaphor Risks Harm During Early Adoption Phase
By
–
I hear you. I'm just worried that this metaphor is increasing the risk of harm in this transition phase when most people aren't familiar yet about what this technology is/isn't (ex https://
x.com/ClementDelangu
e/status/1710349025793159338
…) -
Risks of Making AI Sound Too Human
By
–
I hear you but think I care less about the philosophical debate about the nature of thinking but worry more about the risk of making it sound human like I described here:
-
Anthropomorphism in AI systems: risks and ethical concerns
By
–
How do you think it's useful (genuine question)? For me, it's detrimental because it biases our thinking and perception in giving human attributes to a system that is not leading to things like addiction, ability to control and manipulate and irrational fears like existential
-
Partner with IWF upfront to prevent illegal content in AI datasets
By
–
Partner with IWF upfront, not when problem arise! LAION knew about the risks of illegal content (e.g. child abuse) being in their dataset. It would have been sensible to partner with professional organizations like IWF to check files *before* publishing the dataset.
-
Database Storage for LAION-5B: Removing Illegal Content Safely
By
–
Use a database storage method not version control. Storing the LAION-5B dataset on revision control means nothing can be removed (e.g. private data or illegal content) without leaving a trace or nuking history. Using a database storage + wrapper API would avoid this problem.