Niki Howe (they/them)

About Me

I'm interested in making the future go well for us all in the era of increasingly powerful AI. To this end, I'm currently a research scientist at Meta Superintelligence Labs, working on AI safety for frontier models like Muse Spark.

I completed my PhD at Mila and Université de Montréal, where I worked on AI safety, focusing on topics such as reward hacking, language model robustness, and reasoning model chain of thought monitorability.

Outside of work, I enjoy hiking, meditation, language learning, spending time with friends (and making new ones), and finding serenity amidst the raging storm of conscious experience.

Mentoring

If you are interested in working on mitigating harms from AI and think chatting with me could help, please reach out! I particularly encourage you to do so if you identify as belonging to an underrepresented group in AI safety/ML/CS.

Feedback

Please leave me anonymous feedback here. Since I don't have a way to ask follow-up questions, in case of constructive feedback it's helpful if you give context and/or examples so I can better understand how to apply your feedback in practice.

Rejection

I try to see rejection (and failure in general) as a positive reminder that I'm trying new things and pushing new limits (see growth mindset). Here are some rejections I've experienced.